Chapter 174 - Chapter 167: The Art of Making

Chapter 167: The Art of Making

20 August 1974 — Gorakhpur

The invitation said nine o'clock.

It did not say what the meeting was about.

This was deliberate. When Karan's invitations did not specify the subject, the people who received them understood that the subject was significant enough to resist being reduced to a subject line. They came anyway. They always came anyway, because in four years of receiving invitations from Karan Shergill, the pattern had been consistent: the invitations that said the least were about the things that mattered most.

Forty-one people received the invitation.

They came from across the Gorakhpur complex and from outside it. Chemical engineers from the petroleum division, who had been working on the Jamnagar refinery processes. Process engineers from the steel plants, three of the senior production specialists who had been responsible for the specific manufacturing protocols that had made Shergill Steel's alloy output competitive with German quality. The semiconductor manufacturing team from ISMC — not the research team, the production team, the people who had spent the past two years translating Dr. Chandra's research results into yield curves and wafer throughput and clean room protocols. Two pharmaceutical chemists who had been quietly attached to the Agri-Chem division for four months on an assignment whose formal description was vague. A materials scientist from IIT Bombay who had been consulting on composite manufacturing. A bioprocess engineer from CDRI in Lucknow who had been offered a six-month attachment and had been extended three times.

They also came from further away.

Dr. Prashant Kulkarni had flown from Bombay — he had spent fifteen years at Hoechst's Indian operations before leaving to consult independently, and his specific expertise was the translation of pharmaceutical synthesis from laboratory milligrams to factory tonnes, which was a discipline that did not have a proper academic name in India in 1974 but that Kulkarni had spent fifteen years practicing and that he considered the most important and least understood engineering discipline in industrial chemistry.

Dr. Anjali Srinivasan had taken the overnight train from Madras — she was thirty-eight years old, a fermentation specialist who had spent a decade at the Central Food Technological Research Institute in Mysore working on the specific problems of scaling microbial processes, which was to say: making the bacteria that produced useful things at the scale of a beaker produce them at the scale of a tank, which was not the same operation at a different size but a categorically different engineering problem. She had been contacted two weeks ago by Meera Krishnan and had been given no information beyond: attend this meeting.

Dr. Suresh Pillai had driven from Varanasi — he was fifty-one years old, the head of the Department of Chemical Engineering at BHU, a man who had spent his career watching Indian industry struggle with the specific gap between what laboratories produced and what factories needed, and who had developed strong opinions about why this gap existed and how it could be closed. He had been trying to close it within the constraints of a university department for twenty years. He had not succeeded, because the constraints were too severe. He came to this meeting because the invitation came from a source that did not appear to have the same constraints.

Ranjit Singh came from the Gorakhpur factory floor. He was forty-three years old, a production engineer with no advanced degree and seventeen years of manufacturing experience across three factories, and he had been identified by Meera Krishnan as the person in the complex who had the best practical understanding of why manufacturing processes failed to achieve their theoretical yields. He was not certain why he had been invited to a meeting that appeared to include PhDs from universities and research institutions. He wore his factory uniform because it was his uniform and he did not own a suit.

They gathered in the main conference room in the administration building.

The conference room had been rearranged.

Instead of the standard board-meeting layout — chairs around a rectangular table, a hierarchy implied by the arrangement — the chairs were in a wide oval with no head, no foot, no position that was visually more significant than any other. In the centre of the oval, a small table with a model. The model was covered with a cloth.

Meera Krishnan was at the entrance when people arrived. She directed them to seats without explaining the seating arrangement. She answered no questions about the meeting's subject.

At nine o'clock exactly, Karan came in.

He did not go to the head of the room because there was no head. He stood at the edge of the oval and looked at the forty-one people who had been assembled and said nothing for a moment, the specific pause of a man who is deciding where to begin a thing that has been building inside him for a long time.

Then he said: "I have a question for everyone in this room. I want honest answers. I am not asking to embarrass anyone."

They waited.

"Every person here," he said, "has at some point in their career produced something in a laboratory, or optimised something in a process, or solved a problem in a pilot plant — and then watched the solution fail to become what it should have become. The laboratory result that never reached scale. The process optimisation that improved yield on paper and did nothing in the factory. The pilot plant result that looked like a commercial breakthrough and became a commercial disappointment." He paused. "How many of you have had this experience?"

A moment's hesitation.

Then hands went up.

Not all forty-one. But thirty-six of the forty-one.

"Keep them up," he said.

He looked at the thirty-six hands.

"For each of you," he said, "I want to ask a different question. Not why it failed — I know why it failed, broadly. The question is: was the failure inevitable? Or was the failure a consequence of something specific that was missing?"

He let the hands down.

"Think about it," he said. "Not a general answer. The specific failure you were thinking of when you raised your hand. The specific thing that was missing."

He looked at Ranjit Singh in his factory uniform.

"Ranjit," he said. "You were thinking about something. Tell me what it was."

Ranjit Singh looked at the room. He was a man who was comfortable speaking on a factory floor and not entirely comfortable speaking in rooms with university professors. But Karan had looked at him directly, and Karan's invitations did not allow for declining.

"The extrusion line," he said. "1971. We had a new polymer formulation from the chemistry team. Laboratory results were excellent — tensile strength 40 percent better than what we were running. We tried to scale it. The extrusion parameters that worked at lab scale produced a completely different material at factory scale. Viscosity changed. Processing temperature window narrowed. The formulation that had been perfect at two kilograms failed at two tonnes." He paused. "The chemistry was right. The process engineering was wrong. The chemistry team was not process engineers. The process engineers didn't understand the formulation. There was a gap between them."

"That gap," Karan said. He did not elaborate. He looked at Dr. Kulkarni.

"The synthesis of an antibiotic," Kulkarni said, without being asked. He had understood the format. "Seven-step synthesis. Perfect at bench scale. At pilot scale, the third step produced an impurity at trace level that didn't appear at bench scale because of the mass transfer dynamics. At commercial scale, the trace impurity accumulated to the point where it failed quality specifications. The chemistry was correct. The scale-up process engineering was not in place. We spent eight months redesigning a step that should have been analysed for scale-up sensitivity before the pilot." He paused. "That eight months represented several crore rupees of lost time and the delay of a product that was genuinely needed."

"The impurity," Karan said. "You could have predicted it."

"Yes," Kulkarni said. "With the right tools and the right knowledge of the specific failure modes of that reaction at scale. We didn't have either in-house. We brought in a consultant from Germany eventually. He identified the problem in three days. Three days. We had spent eight months."

"Why was the German consultant necessary?" Karan said.

Kulkarni looked at him. "Because nobody in India had consolidated that knowledge," he said. "The knowledge exists. It's in academic literature, in proprietary industrial practice, in the heads of experienced engineers scattered across different industries. But it's scattered. There's no institution that organises it, that develops it, that produces engineers who are specifically trained in it."

"That," Karan said, "is what we are going to build."

The room was quiet.

He walked to the model on the centre table.

He removed the cloth.

The model was of a building complex — not one building but a cluster of six, arranged around a central shared infrastructure, connected by covered walkways, each building labelled with a small card.

He pointed to the labels in turn.

"Pharmaceutical and bioprocess scale-up division. Chemical process scale-up division. Materials and polymer scale-up division. Semiconductor and electronics manufacturing division. Agricultural and food processing division. Shared analytical and computational infrastructure."

He stepped back.

"The Shergill Process Engineering Institute," he said. "SPEI."

The silence lasted approximately eight seconds.

Then Dr. Pillai from BHU said: "This is not a research institute."

"No," Karan said.

"The purpose is not to discover new knowledge," Pillai said.

"The purpose is to translate existing knowledge into practice," Karan said. "Specifically: to develop and systematise the engineering knowledge that sits between laboratory discovery and commercial production. The knowledge that tells you: when this synthesis works at bench scale and you want to make it work at commercial scale, what do you need to know, what do you need to change, and how do you do it reliably?"

Pillai was quiet for a moment.

"In the academic world," he said, "this knowledge is considered — it is not considered low-status, but it is not where the prestige accumulates. The prestige is in discovery. The prestige is in the novel synthesis, the new molecule, the new material. The translation from discovery to manufacture is treated as an industrial problem rather than a scientific one."

"Yes," Karan said. "And that is precisely why the gap exists. The discovery community treats scale-up as someone else's problem. The manufacturing industry treats it as an engineering challenge that each company solves individually, from scratch, without sharing knowledge." He paused. "The result is that every company that tries to scale a new process rediscovers the same failure modes that dozens of companies before them have already encountered. The knowledge exists, scattered across fifty years of industrial experience in hundreds of companies in dozens of countries. It is never consolidated. It is never systematised. It is never made available to the person who needs it the first time rather than after eight months of expensive failure."

He looked at Ranjit Singh.

"When the polymer extrusion failed," Karan said. "What would have helped you?"

Ranjit Singh thought about it.

"An engineer," he said. "Specifically: an engineer who had scaled polymer formulations before. Who knew what the viscosity sensitivity to temperature looked like at different scales. Who could have told me in advance what the processing window would narrow to before we put the material in the extruder." He paused. "What we had was a materials scientist who was excellent at formulation and a process engineer who was excellent at extrusion but who had no experience of this specific formulation class. We needed someone who was excellent at both simultaneously, or who had encountered this specific combination of problems before."

"That person doesn't exist in one body most of the time," Karan said. "The knowledge exists, distributed across many people. SPEI's job is to consolidate the knowledge, train engineers who hold it, and make it available to any process that needs it."

Dr. Anjali Srinivasan said: "Fermentation." She said it with the specific quality of a person whose entire professional experience has been of this problem and who is hearing it described correctly for the first time. "The gap between a laboratory fermentation and a production fermentation is — it is not one problem. It is fifty problems that interact with each other. The oxygen transfer dynamics change with scale. The mixing patterns change. The heat removal changes. The shear stress on the organisms changes. The dissolved CO2 accumulation changes. None of these changes is individually unpredictable — they can all be modelled, they can all be understood in principle. But the engineer who understands all fifty simultaneously, and who understands how they interact, and who can redesign the process to address them before the scale-up rather than after — that engineer takes years to develop and does not come out of any university programme in India."

"Or anywhere," Karan said.

"Or anywhere," she agreed. "The knowledge is built over careers. Individual careers. Engineers who work their whole lives in one company and accumulate the knowledge and then retire and the knowledge retires with them."

"SPEI captures the knowledge," Karan said. "Systematises it. Makes it institutional rather than individual. Trains the next generation in it explicitly rather than hoping they accumulate it by accident."

"You're describing something that doesn't exist," Srinivasan said. Not as a challenge. As a statement that she was making to confirm she had understood correctly.

"Yes," Karan said.

"Anywhere in the world," she said.

"The closest analogy is the MIT Process Systems Engineering centre," Karan said. "Which is an academic programme. It produces researchers who understand process systems theoretically. The gap between theoretical process systems understanding and practical scale-up knowledge is — substantial." He paused. "What I am describing is not the academic version of this. It is the practical version. The people who work at SPEI will work on real scale-up problems for real products going from real laboratories to real factories. The theoretical knowledge will be developed in the service of practical problems, not in advance of them."

Dr. Kulkarni said: "Who are the clients?"

"Every Shergill division," Karan said. "Petroleum. Steel. Pharmaceuticals — we are entering pharmaceuticals." This was the first time most people in the room heard this. "The LED manufacturing scale-up from the ISMC research programme. The composite materials manufacturing for Shergill Aeronautics. The agricultural processing programmes that are coming from the cold chain expansion. Every one of these has a scale-up problem. Currently each division solves its scale-up problems individually, with whatever resources it has, which are usually insufficient." He paused. "SPEI becomes the shared resource. The institutional knowledge base that every division can access."

"Internal clients only?" Pillai said.

Karan looked at him.

"Not eventually," he said. "In the first three years, the priority is the Shergill divisions' problems. We need to build the knowledge base and the team and the credibility. After three years, SPEI will be available to external clients — Indian pharmaceutical companies, Indian chemical companies, Indian materials manufacturers. Any Indian company with a scale-up problem."

"At what cost?" Kulkarni said.

"At cost for Indian companies below a certain revenue threshold," Karan said. "Commercial rates for the larger ones. The model is not charitable — SPEI is not a charity. But the pricing reflects the specific industrial development objective, which is to close India's scale-up gap broadly, not only for Shergill."

Pillai was quiet for a moment. He was a man who had spent twenty years in a university watching Indian industry struggle. He had many opinions about why Indian industry struggled, and most of those opinions involved the inadequacy of the institutional infrastructure that connected discovery to manufacture. He was sitting in front of someone who was building that infrastructure.

"The faculty," he said. "The people who will work here. Who are they?"

"That," Karan said, "is the most important question." He looked at the model. "The standard approach to building a technical institute is to hire academics. People with PhDs who have published papers. This is the right approach for an institute whose purpose is research and teaching. It is the wrong approach for an institute whose purpose is applied scale-up knowledge."

"You're not hiring academics," Pillai said. He did not sound offended. He sounded interested.

"I am hiring practitioners," Karan said. "People with deep industrial experience in specific scale-up domains. Who have solved real problems in real factories. Who have the specific knowledge that comes only from standing next to a reactor that is not working the way it should and understanding why." He paused. "Some of them will also have academic credentials. Some will not. The credentials are secondary. The knowledge is primary."

He looked around the room.

"Several of the people in this room will be asked to join SPEI," he said. "Not all of you. The ones who will be asked are the ones who have the specific knowledge I've described. You'll know if it's you because Meera will call you this week."

A few people looked at Meera. She was standing at the edge of the room and her expression communicated exactly nothing, which was her characteristic response to being looked at when she was not yet the relevant party in the conversation.

Dr. Kulkarni spoke again. He was fifty-one years old and he had spent his career in the specific intersection of chemistry and manufacturing that SPEI was describing as its domain, and he had spent those years in the condition of a man who is doing something important that no one has named. He was being given a name.

"The pharmaceutical scale-up problem," he said. "Specifically. I want to understand what SPEI does with a pharmaceutical scale-up problem, because this is the domain I know."

"Tell me the problem," Karan said.

"Discovery laboratory produces a synthesis. Five steps, let's say. Works at fifty milligrams. Works at five grams. Now you want to make five hundred kilograms per batch. The steps that work at five grams may or may not work at five hundred kilograms. Some won't. The question is: which steps will fail at scale, and how will they fail, and what do you need to change before you run the first pilot batch rather than after?"

"How do you currently answer that question?" Karan said.

"With experience," Kulkarni said. "Specifically: with the experience of an engineer who has scaled similar reactions before. Who has seen this class of reaction failure before. Who can look at the synthesis route and identify the steps that are likely to have scale-dependent behaviour." He paused. "The problem is that this experience is in people's heads. It is not systematised. There is no database of reaction types and their known scale-up failure modes. There is no decision tree that a less experienced engineer can follow. There is no body of literature that is organised around practical scale-up knowledge rather than discovery chemistry."

"SPEI builds that database," Karan said. "Over the course of solving real problems. Every scale-up problem that SPEI works on — every failure mode identified, every solution developed, every process redesigned — goes into the knowledge base. The knowledge base grows. The next engineer who faces a similar problem has a starting point that is not zero."

"A knowledge commons," Srinivasan said.

"Yes," Karan said. "A knowledge commons for scale-up engineering. Which does not exist."

"Why doesn't it exist?" Pillai said. He was asking as a professor, the specific question of a man who wants to understand the system failure rather than just the symptom.

"Because the knowledge is currently owned by companies," Karan said. "Each company that solves a scale-up problem solves it for themselves and keeps the solution. It is proprietary. The solution for a specific fermentation problem at Company A is A's competitive advantage. They have no incentive to share it. The knowledge accumulates in companies as trade secrets and in individual engineers as experience. Neither is accessible to the next person who faces the same problem."

"SPEI breaks the proprietary model," Kulkarni said.

"SPEI builds a commons that is shared," Karan said. "The specific solutions to specific client problems remain confidential. The underlying engineering principles — the generalised knowledge about how this class of problem behaves at scale — is consolidated and made available. The clients own their specific solutions. SPEI owns the pattern language."

The room was processing this.

Ranjit Singh said: "The polymer extrusion problem I described. What does SPEI do with that?"

"When the extrusion problem arrives at SPEI," Karan said, "the team assigned to it does four things. First: they characterise the failure mode at the scale where it first appeared. Not just 'viscosity changed' but specifically: what was the relationship between scale and viscosity, what were the controlling parameters, what were the measurement signatures that preceded the failure. Second: they search the knowledge base for analogous failures. Has this class of viscosity-temperature coupling been observed before at scale in similar polymer systems? What did those cases look like? What solutions were found? Third: they design a scale-up protocol that tests for the specific failure modes before the commercial run rather than in it. Fourth: they add the problem, the analysis, and the solution to the knowledge base so the next engineer who encounters a similar polymer system has a head start."

Ranjit Singh was quiet.

"The eight months that Kulkarni ji spent," he said. "Would become how long with SPEI?"

"If the knowledge base already contained analogous fermentation impurity cases," Karan said, "the German consultant's three days was the ceiling. With a dedicated team and a knowledge base that was maintained and accessible — possibly one day."

"One day," Ranjit said.

"The German consultant had the knowledge in his head," Karan said. "He had seen this type of impurity accumulation before. He knew what to look for. The three days was him looking. If the knowledge base had the right structure and the right content — the looking time disappears."

Srinivasan said: "The fermentation problem. Oxygen transfer dynamics at scale. This is my specific domain." She was speaking with the quality of a person who has been thinking about this for a long time and is now thinking about it in a new context. "The relationships between scale, mixing power, reactor geometry, and oxygen transfer rate — these are understood. The equations exist. The parameters are measurable. What is not systematised is the specific experience of how these relationships interact in complex biological systems under real process conditions." She paused. "The textbook equations work for simple systems. They are starting points for complex systems. The engineer needs to know: what are the deviations from the theoretical prediction, in what directions do they occur, and what process modifications address them?"

"Which your institute would systematise," Pillai said.

"Which SPEI would systematise," Karan said. "From real cases. From the fermentation problems brought to SPEI by clients. Not from theory."

"Learning by doing," Pillai said.

"Learning by solving," Karan said. "Which is more specific than doing. The doing without the systematic knowledge capture is what currently happens at individual companies. They solve problems. The knowledge stays in their files. SPEI solves problems and extracts the transferable knowledge from each solution."

At ten-thirty, Karan removed four large boards from the wall behind him.

The boards had been covered and were now uncovered.

Each board was labelled with one of the institute's four primary divisions and contained a structured description of what that division did, what knowledge domains it covered, what problems it addressed, and what its interface with the broader ecosystem looked like.

He walked through them.

Pharmaceutical and Bioprocess Scale-Up Division.

The board described the knowledge domains: synthetic organic chemistry scale-up, fermentation process scale-up, downstream processing scale-up (the purification and isolation of biological products from the messy broths that fermentation produced), formulation scale-up (turning an active pharmaceutical ingredient into a stable tablet or injection), and sterile manufacturing process scale-up. The specific failure modes covered under each domain were listed — seventeen for synthetic organic chemistry alone.

He pointed to the synthetic organic chemistry failure mode list.

"Seventeen known failure modes," he said. "Each of them has been encountered in industrial practice in the past thirty years. Each of them has been described somewhere in the technical literature or in proprietary company reports. Each of them has been solved, by someone, somewhere, at least once." He paused. "Not one of them is systematically taught in any Indian university chemical engineering programme. Not one of them is covered in the standard textbooks used in those programmes. The student who graduates from IIT or Bombay University with a chemical engineering degree has no systematic knowledge of any of these failure modes."

"The ISMC programme," Dr. Ashwini Nagarkar from Pune said — he was a chemical engineer who had been working on the pharmaceutical scale-up problems and had been listening with the close attention of a man who was recognising his own experience being described back to him. "The amoxicillin process. The impurity problem. Which of the seventeen does that fall under?"

"Four," Karan said. "Mass transfer limited side reaction. Intermediate accumulation. Heat of reaction mismatch at scale. Impurity profile change with residence time distribution." He pointed to numbers three, seven, eleven, and fourteen on the board. "At bench scale, the residence time distribution in a stirred flask is narrow — all the material experiences similar reaction times. At pilot scale, the residence time distribution broadens. The material that experiences longer residence times undergoes a secondary reaction that doesn't occur at the shorter times in the flask." He paused. "Understanding this mechanism requires knowing that residence time distribution changes with scale, knowing how to measure it, knowing which reaction classes are sensitive to it, and knowing what process modifications address it. All of this knowledge exists. None of it was available to the person who spent eight months finding it the hard way."

Nagarkar was quiet.

Then he said: "I spent six months on a similar problem in 1971. Completely independently. I arrived at a similar solution through a different route. If there had been a resource — a systematic record of this failure mode — I would have solved it in weeks."

"Yes," Karan said.

"And the resource doesn't exist," Nagarkar said.

"Everywhere in the world," Karan said. "Individual companies have their internal knowledge. It is not shared."

"You are creating a shared knowledge institution," Nagarkar said.

"Yes," Karan said. "With the specific structure required to make the sharing work. The knowledge is organised by failure mode class, not by product or company. The companies keep their specific product knowledge. SPEI holds the generalised failure mode knowledge."

He moved to the next board.

Chemical Process Scale-Up Division.

The domains: heterogeneous catalysis scale-up, continuous versus batch process conversion, solvent recovery and recycling at scale, heat integration at scale, materials handling for viscous and solid-containing streams. The failure modes listed here were different from the pharmaceutical ones but shared the same structural characteristic: known, documented somewhere, not systematically available to the engineer who needed them.

He pointed to a specific failure mode: catalyst deactivation pattern change with reactor scale.

"The Jamnagar hydrocracking unit," he said. "We commissioned it earlier this year. The catalyst activity profile in the commercial reactor was different from the pilot plant profile. Expected — catalyst deactivation is scale-dependent for reasons that are understood theoretically but that produce specific patterns in specific catalyst systems that require empirical characterisation." He paused. "We had the right engineering team and the right monitoring systems and we managed the difference correctly. It cost us three weeks of suboptimal performance during initial operation. With the systematic knowledge that SPEI would hold, based on the documented catalyst deactivation patterns in similar systems — we would have predicted the specific pattern, designed for it in advance, and the three weeks disappears."

"Three weeks of suboptimal performance on a hydrocracker," the petroleum division engineer in the room said. "At Jamnagar throughput, that's—"

"Significant," Karan said. "Multiply that by every scale-up transition across every Shergill division in a year. Every transition has a suboptimal period that systematic scale-up knowledge could shorten. The aggregate saving is substantial." He paused. "But the number is not the point. The number is the commercial case for SPEI. The actual point is what it means for Indian manufacturing broadly — for every Indian company that is currently spending months solving problems that SPEI would solve in days."

Materials and Polymer Scale-Up Division.

The domains: composite material manufacturing scale-up, polymer processing scale-up, ceramic manufacturing scale-up, advanced alloy processing scale-up. He spent longer on this board than the others.

"India is developing indigenous composite materials manufacturing," he said. "The Arjuna armour panels. The S-35 airframe composite sections. These processes were developed at laboratory and pilot scale. They are now in production scale-up. The specific failure modes in composite manufacturing at scale — void formation, fibre alignment variation, matrix distribution inhomogeneity, cure cycle deviation effects — these are well-documented in the aerospace manufacturing literature. The literature is in English, in journals that are not widely available in India, in technical standards documents that are classified or proprietary." He paused. "The SPEI materials division compiles, organises, and makes available the systematic knowledge of how composite manufacturing processes fail and succeed at scale. Not in six months when the problem appears. Before the process is designed."

He pointed to a specific case study on the board.

"Carbon fibre composite panel manufacturing," he said. "The transition from two-metre panels to four-metre panels. The cure cycle that works perfectly for a two-metre panel produces warping in a four-metre panel because the thermal mass of the larger panel creates a temperature gradient during cure that the smaller panel does not experience. The warping is predictable — the relationship between panel size, thermal mass, and temperature gradient during cure is understood. The corrective action is known — modified cure profile with controlled ramp rate. This knowledge is in the literature." He paused. "It is in a 1968 Boeing technical report. It is not in any Indian materials engineering textbook. The Indian engineer who doesn't know the Boeing report will discover the problem the first time they scale to four metres. The SPEI engineer knows the Boeing report. And two hundred other documents like it."

Subramaniam from the aeronautics composites division said: "We had the warping problem."

"I know," Karan said.

"We solved it in three months," Subramaniam said. "Empirically. Trial and error."

"With the SPEI knowledge base: three days," Karan said. "The solution was known in 1968. The knowledge was not accessible."

Subramaniam was quiet.

Semiconductor and Electronics Manufacturing Division.

"This division exists primarily to support ISMC's scale-up from research and pilot production to commercial manufacturing," Karan said. "The blue LED is confirmed. The commercial production programme begins this year. The scale-up from three-inch wafers to six-inch wafers, from manual phosphor dispensing to automated dispensing, from research-grade reactor conditions to production-grade conditions — these transitions have well-documented failure modes in the semiconductor manufacturing literature. The SPEI semiconductor division holds that knowledge and applies it to the ISMC transition."

He paused.

"It also holds the knowledge for future semiconductor technologies," he said. "The next generation of ISMC products — when those are ready for production scale-up — the knowledge base is already there. The engineers are already trained. The transition time shrinks with each cycle because the institutional learning accumulates."

He stepped back from the boards.

"One more thing," he said. "The shared infrastructure."

He pointed to the sixth building on the model.

"Analytical and computational infrastructure," he said. "Scale-up engineering is increasingly a computational discipline. The ability to model how a process will behave at scale before building the scale — to simulate the heat transfer, the mass transfer, the fluid dynamics, the reaction kinetics — is the difference between designing a scale-up process and guessing one. The computational infrastructure at SPEI will be available to all six divisions. The analytical equipment — the measurement systems for characterising processes at scale — will be shared." He paused. "Shared infrastructure is why SPEI works as a single institute rather than as five separate division programmes. The fermentation engineer and the polymer engineer use the same modelling software, the same reactor characterisation equipment, the same data analysis tools. The knowledge transfer between domains happens because the people are in the same building using the same infrastructure solving structurally analogous problems."

Dr. Pillai from BHU had been listening with the sustained attention of a man who was simultaneously evaluating an institutional concept and revising his model of what was possible within a private industrial context. He said:

"I want to raise a structural question. Not a critical question — a structural one."

"Raise it," Karan said.

"The knowledge base," Pillai said. "The systematic compilation of failure modes and solutions. Building that knowledge base requires that the engineers who solve problems at SPEI are also producing documented, generalised knowledge from those solutions. That is a different skill set from either laboratory research or industrial engineering. The researcher who solves a problem in a factory usually documents the specific solution for the specific product. The academic who writes a review paper documents the general knowledge but may not have solved the specific problem. SPEI needs people who can do both — solve the specific problem and extract the general principle."

"Yes," Karan said. "That is the hardest staffing problem."

"How do you solve it?" Pillai said.

"By training for it explicitly," Karan said. "SPEI will have a training programme — not a degree programme initially, but a structured development programme for the engineers who work here. The programme will teach both the domain knowledge and the knowledge capture methodology. The documentation standards, the failure mode taxonomy, the generalisation protocols. The engineers will learn not just to solve problems but to extract and record the transferable knowledge from each solution."

"The methodology for extracting generalised knowledge from specific solutions," Pillai said. He was thinking about this. "That methodology itself is not well-developed. The formal epistemology of how you go from a specific case to a general principle that is applicable to analogous cases — that is an active research area in engineering epistemology."

"We will develop it here," Karan said.

"By doing it," Pillai said.

"By doing it," Karan confirmed. "The methodology for knowledge capture is itself a scale-up problem. We start with the best current understanding, apply it, find where it fails, improve it. The knowledge capture methodology improves with each generation of problems."

Pillai looked at the model. He looked at the boards. He looked at the forty other people in the room.

"This is not an engineering institute," he said. "Not in the conventional sense."

"No," Karan agreed.

"It is a meta-engineering institute," Pillai said. "Its subject is not any engineering domain. Its subject is the process of engineering knowledge transfer between scales. It is an institute about how engineering knowledge moves from small to large."

"Yes," Karan said. "That is the correct description."

"Has anyone built this before?" Pillai said. He was asking with genuine curiosity, not with the rhetorical question intonation of someone who already knows the answer is no.

"Not with this specific purpose," Karan said. "There are industrial research institutes that work on manufacturing problems. There are university process systems engineering programmes. There are corporate R&D centres that focus on scale-up. None of them has the specific structure I am describing: a shared, cross-domain, explicitly knowledge-capturing institution whose primary product is not solutions to specific problems but the organised knowledge that makes future solutions faster."

"You are inventing a new kind of institution," Pillai said.

"Yes," Karan said. And then, with the specific directness of someone making a statement they have thought about carefully and are prepared to stand behind: "India needs it. The gap between what India can discover and what India can manufacture is not a discovery capability gap. India's scientists and engineers are as capable as any in the world. The gap is in the infrastructure for translating discovery into manufacture. SPEI is that infrastructure."

The break at eleven-thirty was not a break so much as the continuation of the meeting in smaller groups, the formal structure dissolving into the informal structure that significant meetings produced when people who had been listening to the same thing for two and a half hours found themselves with similar questions and different experiences and the need to process both.

Kulkarni and Srinivasan found themselves at the tea table at the same moment.

Kulkarni poured his tea. Srinivasan poured hers. They stood together at the table looking out the window at the Gorakhpur complex — the factory buildings, the construction cranes of the expansion, the specific landscape of a large industrial complex in operation.

"Fifteen years," Kulkarni said.

Srinivasan looked at him.

"Fifteen years I have been telling people that this gap exists," he said. "The pharmaceutical industry, the government, the academic institutions. Every time I describe what is missing, people agree that it is missing. Nobody builds it."

"Because nobody has the resources and the motivation simultaneously," Srinivasan said.

"And he has both," Kulkarni said, with a slight gesture toward the room where Karan was continuing a conversation with Pillai and two other engineers.

"The resources are obvious," Srinivasan said. "The motivation is what interests me more. Most industrialists who build institutions build them for their own commercial purposes. This is broader than his commercial purposes."

"Is it?" Kulkarni said.

"The external clients," she said. "The availability to Indian industry broadly. He could have built a captive internal scale-up capability for Shergill divisions and stopped there. The commercial case for that would have been strong. Instead he's building an institution that is explicitly designed to benefit Indian industry in general."

"He said it," Kulkarni said. "Directly. The scale-up gap is an Indian industrial gap, not just a Shergill gap. He's solving the Indian gap."

"Why?" she said.

Kulkarni thought about it.

"Because he is the kind of person who, when he sees a problem, solves the problem," he said. "Not the surface problem. The structural problem." He drank his tea. "The surface problem is: my pharmaceutical division has scale-up difficulties. The structural problem is: India has no systematic scale-up infrastructure. He is solving the structural problem because that is what needs solving."

Srinivasan was quiet.

"The fermentation division," she said.

"Yes?" he said.

"He wants me to lead it," she said. "Meera Krishnan called me three days ago. The meeting was described as an exploration. Meera doesn't do explorations. She does confirmations."

"And?" he said.

"And I've been in Madras for fifteen years," she said. "I have a good programme at CFTRI. I have students. I have colleagues."

"You also have a problem," he said. "The specific problem that SPEI is designed to address. The knowledge you've accumulated about fermentation scale-up — it lives in your head. In your lab notebooks. In the papers you've published, which are read by a few hundred people. In the patents your institute holds, which are not being exploited." She looked at him. "He is offering you a platform that matches your knowledge with the resource base to do something with it."

"At the cost of leaving Madras," she said.

"At the cost of leaving Madras," he agreed. "Which is a real cost." He paused. "What is the benefit?"

She looked at the window.

"The benefit," she said, "is that the knowledge I've accumulated over fifteen years stops being only mine."

When the meeting reconvened after the break, the quality in the room had changed.

The forty-one people had been processing for forty-five minutes. Some of them had been processing what they had heard and finding it significant in the way significant things were found significant: not with excitement, which was the first response, but with the specific gravity of understanding what something meant for the work they had been doing and the work they would do.

Karan stood at the model again.

"I want to talk about the timeline," he said. "Because the timeline is not the normal timeline for building an institution."

He looked at the room.

"Most institutions are built in phases over several years," he said. "The first phase is planning. The second phase is construction. The third phase is recruitment. The fourth phase is programme development. The fifth phase is first operations." He paused. "SPEI's timeline is different. The first problem is already here."

He looked at Dr. Chandra's representative in the room — a young engineer from the ISMC production team named Sunil who had been sent to the meeting because Chandra himself was in the lab, which was where Chandra was always.

"The blue LED production scale-up," Karan said to Sunil. "The transition from three-inch wafer production to six-inch wafer production. Current status?"

Sunil straightened slightly. "The six-inch wafer reactor modification is complete," he said. "We have three runs with the six-inch substrate. The yield is 68 percent on the target efficiency specifications. We expected 75 percent. The gap is primarily in the uniformity of the InGaN active layer across the larger wafer area."

"What is causing the uniformity issue?" Karan said.

"Temperature gradient across the wafer during growth," Sunil said. "The three-inch reactor had a temperature uniformity of plus or minus two degrees across the substrate. The six-inch reactor has plus or minus five degrees. We don't yet know how to close that gap."

"How long to solve it empirically?" Karan said.

"Three months minimum," Sunil said. "Possibly six. We're doing a systematic temperature profiling study."

"That problem," Karan said to the room, "is a classic MOCVD scale-up problem. The temperature uniformity challenge in larger reactors. It is documented in the semiconductor manufacturing literature going back to the 1960s for gallium arsenide systems. The solutions are known — susceptor design modification, rotation rate optimisation, gas flow geometry adjustment. The specific application to InGaN/GaN requires calibration for our material system, but the framework for the solution is available."

He looked at Sunil.

"The SPEI semiconductor division solves this in three weeks," he said. "Not by discovery. By application of documented knowledge from analogous systems to our specific situation."

He looked at the room.

"That is the first SPEI project," he said. "It begins the day SPEI has a functioning semiconductor team. Which is the first working day after this meeting."

The room absorbed this.

"You are opening with an active problem," Pillai said.

"The first project is the construction of the knowledge base and the validation of the methodology simultaneously," Karan said. "SPEI does not build the knowledge base in theory and then apply it in practice. It builds it by applying it. The first project is real. The next project is real. The knowledge base builds from real solutions to real problems."

"The building isn't constructed yet," Kulkarni observed. He was looking at the model.

"The building breaks ground today," Karan said. "Construction completes in four months. In the interim, the SPEI teams operate in the existing ISMC and engineering division spaces. The work starts today. The building follows the work."

"You are building the institution while it operates," Pillai said.

"I am building the institution by operating it," Karan said. "The institution is its work. The building is the shelter for the work."

At one o'clock, the model was moved to the side and lunch was brought in. Not catered — the factory canteen's food, in the large containers that fed the factory floor workers. Karan had made this specific choice deliberately and had told Meera Krishnan to make it when the catering was being organised.

The food arrived in the containers. The containers were placed on a long table. People served themselves.

It was the specific quality of a meal in a working institution rather than a formal occasion, and it produced the specific quality of conversation that institutional meals produced: the conversation that was not the meeting, which was often more significant than the meeting.

Ranjit Singh found himself at a table with Dr. Pillai and Dr. Srinivasan and a young engineer from the composites division named Karthik.

Ranjit Singh had been thinking about the polymer extrusion problem since he named it that morning.

"The three months we spent on that problem," he said. He was talking to the table in general, or to himself, the quality of a man continuing an internal conversation. "I remember every day of it. The frustration. The runs that were wrong in ways we didn't understand. The chemistry team arguing with the process team. The management pressure." He paused. "And then we solved it. Not elegantly — empirically, by trying enough things until one of them worked. And we never wrote it down properly. The solution is in a maintenance log somewhere. Not documented as a failure mode analysis. Not in a form that any engineer who faced the same problem later could find and use."

"That is the specific structural failure," Pillai said. "The problem is solved. The knowledge is not captured. The next engineer starts from zero."

"If SPEI had existed," Ranjit said. He was doing the arithmetic. "If the failure mode was in the knowledge base. We call SPEI. They pull up analogous cases. They give us a starting point. We don't spend three months. We spend maybe two weeks."

"Less," Srinivasan said. "With the right knowledge base and the right analytical support, the diagnosis phase for a known failure mode class is days, not weeks. The optimisation phase — finding the specific parameters for your specific system — takes longer. But you're optimising from a known direction rather than searching in the dark."

"The diagnosis is the expensive part," Karthik said. "In composites, the cure cycle warping problem — we spent six weeks just figuring out that it was a thermal gradient issue. Once we knew it was thermal gradient, the solution took two weeks. But six weeks to identify the root cause."

"With the SPEI knowledge base," Karan said. He had come to their table. He sat down with his own plate. He did not sit at a separate table. "The diagnosis is in the database. Cure cycle warping in large composite panels — thermal gradient during cure is the primary cause in over 80 percent of documented cases. The database tells you: look at the thermal profile first. Check the cure cycle ramp rate. Measure the temperature differential across the panel. The six weeks of diagnosis becomes: check the thermal profile on day one."

Karthik looked at him. "We should have known to check the thermal profile on day one."

"Yes," Karan said. "You should have. The knowledge that thermal gradient was the primary cause was available in the literature. It wasn't available to you because the literature is not organised for your use case. SPEI organises it."

Ranjit Singh was quiet for a moment.

"Mr. Shergill," he said.

"Yes," Karan said.

"I've been on the factory floor for seventeen years," he said. "I have no PhD. I have no publications. I am an engineer by experience rather than by certification." He paused. "Is there a place in SPEI for someone like me?"

Karan looked at him.

"Ranjit," he said. "The most valuable knowledge in SPEI's knowledge base is the knowledge that comes from seventeen years on the factory floor. The knowledge of how processes actually fail, not how they are theoretically supposed to fail. The knowledge that lives in your hands and your eyes and your nose—" he meant this literally, the specific sensory experience of a man who had spent seventeen years knowing when a process was wrong by how it smelled or sounded before the instruments registered it "—that knowledge is the foundation that the theoretical framework sits on. Without it, the knowledge base is incomplete. The theoretical framework without the practical knowledge produces engineers who understand why processes fail without being able to recognize when they are about to fail."

Ranjit Singh absorbed this.

"There is a role for you at SPEI," Karan said. "Not an administrative role. Not a management role. A technical role. What you know needs to be captured and systematised. The knowledge in your head — the seventeen years of pattern recognition and failure mode experience — is exactly what the knowledge base needs."

"I've never written a technical document in my life," Ranjit said.

"SPEI will teach you to," Karan said. "The knowledge capture methodology that Pillai ji described this morning — that is a learnable skill. You have the underlying knowledge. The methodology sits on top of it."

Ranjit Singh looked at his food.

"I'll think about it," he said.

"Think about it tonight," Karan said. "Meera will call you tomorrow."

The afternoon session was different from the morning.

The morning had been Karan describing SPEI. The afternoon was the people in the room examining it — the specific examination that a group of forty-one people with different expertise and different experience applied to something they had been thinking about for three hours.

The questions were good.

Dr. Nagarkar from Pune said: "The proprietary knowledge problem. When SPEI solves a client's problem, the solution belongs to the client. But the generalised failure mode knowledge goes into the shared base. How do you prevent the generalised knowledge from being reverse-engineered into the specific solution?"

"The taxonomy," Karan said. "The knowledge base is organised at the level of failure mode class, not at the level of specific product or process. The entry for 'intermediate accumulation in multi-step synthesis' describes the mechanism, the diagnostic indicators, the known solution approaches, the optimisation parameters. It does not describe which specific intermediate in which specific synthesis accumulates at what specific scale for which specific client." He paused. "The specificity that would allow reverse-engineering to a client's product is not in the shared base. It is in the client's own records."

"Is that distinction always maintainable?" Nagarkar pressed.

"Not always," Karan said. "There will be cases where the failure mode description is specific enough that a sophisticated reader could infer the client's situation. In those cases, the entry in the knowledge base is written at a higher level of abstraction. The team member responsible for knowledge capture makes the judgment about the appropriate level of abstraction. This is a skill that requires development. It is part of the training."

"The judgment call," Nagarkar said. "That's a human judgment that can go wrong."

"Yes," Karan said. "Every system that involves human judgment can go wrong. The alternative is to make the knowledge base so abstract that it loses practical value. That is also a failure mode." He paused. "We will make judgment errors. We will learn from them and improve the taxonomy. This is the same process as every other aspect of SPEI — we start with the best current understanding and improve it."

A bioprocess engineer from the agricultural division said: "The agricultural processing domain. You listed it as one of the five divisions. But agricultural processing scale-up is — it's not a domain with a clean body of literature. The failure modes are more variable, more dependent on raw material variability, season, geography."

"Yes," Karan said. "It is the hardest of the five domains for systematic knowledge capture. The variability is genuine and it is large." He paused. "That does not make systematic knowledge impossible. It makes the knowledge base more probabilistic. Instead of 'this failure mode occurs in these conditions,' the agricultural processing knowledge base says 'this failure mode is more likely in these conditions, with these indicators.' The knowledge is less deterministic. It is still more valuable than no systematic knowledge."

"Acknowledging uncertainty explicitly," Pillai said.

"Representing it accurately," Karan said. "Which is what good engineering knowledge always does. The pretense of certainty in domains that are genuinely uncertain is more dangerous than the explicit acknowledgement of uncertainty."

At three in the afternoon, a specific question arrived from an unexpected source.

Sunil, the young ISMC production engineer who had described the six-inch wafer yield problem, said: "The ISMC semiconductor division of SPEI. You said the team starts work immediately on the temperature uniformity problem. Who is the team?"

"Three people initially," Karan said. "A MOCVD specialist from the LED research team who will transition to SPEI. An equipment engineer from the reactor manufacturer who has agreed to a six-month technical secondment. And an external consultant — someone who has worked on MOCVD temperature uniformity at commercial scale elsewhere."

"Elsewhere meaning outside India," Sunil said.

"Yes," Karan said.

"The knowledge we're trying to build domestically," Sunil said, "is being seeded by someone from outside India."

"Yes," Karan said. "That is how knowledge transfer works. You bring in the knowledge that exists elsewhere. You apply it here. You add to it here. In five years, the knowledge base holds knowledge that does not exist elsewhere because it was built from problems that occurred in Indian manufacturing contexts which are specific to India." He paused. "The technology transfer is the beginning. The domestic knowledge creation is the endpoint. The institution is the mechanism that converts the imported seed into the domestically grown crop."

"The wheat analogy," Pillai said.

"Yes," Karan said.

"You import the seed variety," Pillai said. "You grow it here. You select from the crop for the next season's seed. Eventually the variety is adapted to Indian conditions."

"Yes," Karan said.

"The Green Revolution," Pillai said.

"The Green Revolution was knowledge transfer at agricultural scale," Karan said. "Norman Borlaug brought wheat varieties. India grew them, adapted them, bred from them. The HYV wheat India grows now is not Borlaug's original variety. It is a variety that developed from Borlaug's variety under Indian conditions." He paused. "SPEI does this for process engineering knowledge. The imported seed is the existing body of scale-up engineering knowledge. The Indian conditions are the specific manufacturing contexts, the specific raw material variability, the specific infrastructure constraints, the specific workforce characteristics of Indian industry. The crop that SPEI grows is knowledge that is specifically adapted to Indian conditions."

The room was quiet.

At four-thirty, Karan stood at the model one final time.

The forty-one people had been in the room for seven and a half hours. The quality of attention in the room was the specific quality of sustained engagement — not the attention of people who were enduring a long meeting, the attention of people who had been thinking hard about something significant and who were now at the point where the thinking was producing decisions.

"I want to be direct about what I am asking," Karan said.

"Some of you are going to be asked to work here. Not to consult. Not to advise. To come here and work. To relocate to Gorakhpur if you are not already here. To make the specific commitment that building a new institution requires." He paused. "I want to be honest about what that commitment means. It means uncertainty — SPEI is new, its methods are untested, its outcomes are not guaranteed. It means distance from your existing institutions and networks and in some cases from your families. It means doing something that has not been done before, which means that the risk of failure is real."

He looked at the room.

"I am not offering you a safe career choice," he said. "I am offering you something different. I am offering you the chance to build an institution that India needs and that does not exist. The chance to do the kind of work that your expertise has been preparing you to do without the infrastructure to do it. The chance to capture the knowledge that lives in your head and make it available to the next generation of engineers so that they don't spend eight months solving problems that were solved twenty years ago."

He paused.

"That work," he said, "is the most important engineering work in India right now. Not the most glamorous. The most important. Because the gap between discovery and manufacture is the gap that determines whether India's scientific and engineering capability translates into industrial output or remains in the laboratory." He paused. "SPEI closes that gap. I am asking you to come help close it."

The room was quiet for a long moment.

Then Ranjit Singh said: "I'm in."

He said it in Hindi, which was his language, which was the factory floor's language, and it was the first Hindi anyone had spoken in seven and a half hours of meeting. And the directness of it — the specific directness of a man with seventeen years of factory floor experience who had heard what he needed to hear and was done thinking about it — produced something in the room. A quality of reality. The kind of reality that a meeting acquired when someone said the plain thing without elaboration.

Srinivasan said: "I need to speak with my institute director and my husband. If those conversations go as I think they will, yes."

Kulkarni said: "The pharmaceutical scale-up division. If I lead it, I want to bring two people from my current practice."

"Name them," Karan said.

Kulkarni named them.

"Done," Karan said.

Pillai said: "I am retiring from BHU at the end of this academic year. In August 1975 I am available." He paused. "I want to lead the educational programme. Not any of the five divisions. The knowledge capture methodology training. The programme that teaches engineers how to extract generalised knowledge from specific solutions."

"That is the programme I most need to exist," Karan said. "Yes."

Nagarkar said: "My current contract with the pharmaceutical client ends in November. If SPEI exists in November, I will be available."

"SPEI will exist in November," Karan said.

The meeting ended at five-fifteen.

Karan was in his office at eight that evening.

The meeting had been seven and a half hours. He had been awake since five in the morning. He was not tired in the way that long days usually produced tiredness. He was in the specific state of alertness that followed a day when something that had been in the planning stage had moved into the real stage — when the thing that existed in the notebooks and the models and the conversations had become, through the act of describing it to the people who would build it, something that was now also in other people's heads and therefore no longer only in his.

He was thinking about what Pillai had said.

This is a meta-engineering institute. Its subject is not any engineering domain. Its subject is the process of engineering knowledge transfer between scales.

This was accurate.

He had not described it that way to himself. He had described it to himself as: the institution that closes the gap between discovery and manufacture. The gap that every Indian company fell into. The gap that cost months and crore and opportunity and confidence.

But Pillai's description was more precise.

The subject was knowledge transfer. Not knowledge creation — there were institutions for that. Not knowledge application — companies did that. The specific, undervalued, un-institutionalised act of moving knowledge from the scale where it was discovered to the scale where it was useful.

This act was not glamorous. It was not Nobel Prize territory. It did not produce discoveries that were announced on front pages.

It produced factories that worked.

It produced products that reached people.

It produced the specific and unremarkable miracle of an invention that had been discovered in a laboratory ten years ago being available in a chemist shop in Lucknow.

The world was full of inventions that had not made that journey. The laboratories of every country contained inventions that had never become products because the gap between the laboratory and the factory was too wide and too expensive to cross without the systematic knowledge that should have existed and didn't.

SPEI was going to build that knowledge.

Not quickly. Not without failure. But systematically, from real problems, and with the specific institutional structure that made the knowledge transferable rather than trapped in individual heads.

He opened the notebook.

He wrote:

August 20, 1974. SPEI founding meeting. Forty-one people.

Confirmed: Ranjit Singh. Kulkarni (with two staff). Srinivasan (conditional). Nagarkar (November). Pillai (August 1975).

Ground breaks tomorrow. Building complete by December. First operations: immediately.

The first problem is the ISMC six-inch wafer temperature uniformity. Three weeks to solution.

He looked at what he had written.

Then he added:

Pillai described SPEI correctly: a meta-engineering institute. The subject is not engineering. The subject is how engineering knowledge moves from small to large.

Every other Shergill programme produces a thing. The aircraft. The oil. The tank. The LED. The film.

SPEI produces the capacity to produce things. Not a thing, but the capacity to translate things from what they are in laboratories into what they need to be in the world.

This is the most important programme.

Not the most dramatic. The most important.

He thought about this for a moment. Then wrote:

The S-27 makes India safer. The LED makes India smarter. SPEI makes India capable of making its inventions into its industries.

That is the third leg. Discovery. Proof. Scale.

Discovery: the laboratories, the universities, the research programmes.

Proof: the Shergill programmes that build working prototypes.

Scale: SPEI.

Without the third leg, the table falls. India discovers and proves and does not manufacture. The LED is invented in Gorakhpur and manufactured in Japan.

With SPEI: the LED is manufactured in Gorakhpur.

He closed the notebook.

He turned off the desk lamp.

Outside: the factory running the evening shift. The ISMC facility with its lights on, the reactors running, the three-inch to six-inch transition in progress with its 68 percent yield that was going to be 90 percent in three weeks when the SPEI semiconductor team brought the documented knowledge from the analogous cases in the literature to the specific parameters of the Gorakhpur reactor.

Three weeks.

Not three months. Not six months. Three weeks.

That was the value of the institution. Not the millions of rupees in licensing revenue. Not the patents. Not the prestige.

Three weeks instead of six months.

Multiplied by every scale-up problem in every Shergill division.

Multiplied by every Indian company that used SPEI's services over the next twenty years.

The aggregate was not calculable. It was the specific aggregate of a country that was no longer leaving inventions in its laboratories because it lacked the systematic knowledge to build them into products.

He stood up.

He walked to the window.

The factory below him. The lights.

India was a country that could now test nuclear weapons and invent LEDs and build fighter aircraft. It was becoming a country that could also make things — make them at scale, make them reliably, make them efficiently.

That becoming was the work of SPEI.

It was not the work of a year. It was the work of a decade, and the decade started today.

He went home.

End of Chapter 167

Shergill Process Engineering Institute — Founding RecordDate: 20 August 1974Location: Gorakhpur, Uttar Pradesh

Purpose: To systematise the engineering knowledge required for translating laboratory-scale discoveries and processes into commercial-scale production. To build the institutional knowledge base that closes India's scale-up gap.

Divisions:

Pharmaceutical and Bioprocess Scale-Up Chemical Process Scale-Up Materials and Polymer Scale-Up Semiconductor and Electronics Manufacturing Agricultural and Food Processing Scale-Up Shared Analytical and Computational Infrastructure

First Project: ISMC blue LED six-inch wafer temperature uniformity optimisation. Target: 90% yield within three weeks. Current yield: 68%.

Confirmed Founding Team (August 1974):

Ranjit Singh — Production Engineering (factory floor knowledge capture) Dr. Prashant Kulkarni — Pharmaceutical Scale-Up Division (with two associates) Dr. Anjali Srinivasan — Bioprocess Division (conditional on institutional clearance) Dr. Ashwini Nagarkar — Pharmaceutical/Chemical Division (November 2024) Prof. Suresh Pillai — Educational Programme Director (August 1975)

Construction: Ground broken August 21, 1974. Target completion: December 1974.

Interim operations: ISMC and engineering division facilities from August 21, 1974.

Mission statement (Karan Shergill, August 20, 1974):The gap between what India can discover and what India can manufacture is not a discovery capability gap. It is an infrastructure gap. SPEI is that infrastructure.