Six programmes, an innovation centre and a national skills pipeline run on one stack — fab-grade hardware at the edge, simulation and AI in the middle, and a named human accountable at the end.
Bottom is a real wafer in a learner’s hand. Top is a person who signs off what the system produced. Everything else sits between those two facts.
The layer with a person in it. Every lesson, assessment and safety instruction is signed off by a named engineer or educator before a learner sees it, and what we hold about a learner is minimised, disclosed and deletable on request.
Where people meet the system: WOW Bot for questions, the AI Career Compass for direction, the Micro-Learning Studio for teachers, and partner dashboards for institutions and sponsors reporting on their cohorts.
Retrieval-grounded reasoning, task-scoped agents and an orchestration layer above them that routes work, retries what fails and escalates anything uncertain to a human instead of guessing.
The models that score, sort and predict rather than write: learner-progress scoring, career matching, process and yield simulation, and the analytics behind our published impact numbers.
One governed corpus underneath everything — curriculum, equipment documentation, cleanroom protocol, safety data sheets and the WA-01 to WA-08 taxonomy. Retrieval decides what the system is allowed to talk about, which is how a confident wrong answer never reaches a sixteen-year-old.
Digital twins of fab processes, an eight-stage virtual fab, yield and green-fab simulators, AR and VR walkthroughs, and interactive exhibits. This is where a learner is allowed to break things.
The layer that makes the rest credible: real wafers, real process equipment, real metrology, a mobile cleanroom module and instrumented lab benches — carried to campuses on the Fab On Wheels bus.
Parts of this stack run in production today; parts are still in build — the intelligence and interface layers in particular. We do not describe a capability as live until learners are actually using it, and we will tell you exactly where any layer stands if you ask. Our privacy policy →
The same backbone, expressed differently in each programme.
The mobile fab lab — the edge of the whole platform.
The knowledge museum — eighty years of silicon, made walkable.
Learning, assessment and certification across the WA-01 to WA-08 domains.
The humanoid educator and the guide interface behind it.
Turning learners into founders, with the bench to prove an idea on.
Sustainable manufacturing, modelled rather than asserted.
Where research, prototyping and localisation actually happen.
A working floor, not a showroom. Research, prototyping, localisation and student ventures share one set of benches — which is the point, because that is how ideas actually cross between them.
Instrumented workstations for materials characterisation, process measurement and failure analysis at a scale a college can actually host.
Design, fabricate, populate and debug boards — the fastest way for a student to go from an idea to something that switches on.
A working automation cell — controllers, actuators, vision and safety interlocks — so Industry 4.0 is something learners wire up rather than read about.
Local compute for process simulation, yield analytics and model work, so datasets that should not leave the building do not have to.
3D printing, machining, jigs and fixtures — the unglamorous half of turning a prototype into something that survives a demo.
Bench space, mentoring and industry problem statements for Semipreneur teams, so a student venture starts with a real customer instead of a pitch deck.
Not “AI-powered” as a label. Six specific swaps, each of which changes what a learner ends up able to do.
Progress scoring watches where a learner actually struggles and re-sequences what comes next, so a slow start on vacuum theory does not become a permanent ceiling.
Simulation lets a learner destroy a virtual batch, blow a thermal budget and hunt the yield loss — consequences without cost, which is the only way most people learn process discipline.
Assessment is scored against what the learner did in the simulator and on the bench, not only what they ticked, and the model does the first pass so a human reviewer sees the exceptions.
The AI Career Compass maps what a learner enjoys and is good at onto specific fab, ATMP and facilities roles, with the domain codes and tracks that lead there.
Faculty tools turn a lesson plan into lab sheets, a manual into a micro-lesson and a topic into a quiz, so the scarce resource — an experienced trainer — spends time teaching, not formatting.
Content is built to work offline on the bus and on low-end devices, with multilingual delivery, because the learners furthest from a fab are also furthest from good connectivity.
Every request runs the same five steps. The last one is a person.
Research here means something a partner plant, a polytechnic or a student venture can use within a year — not a paper nobody implements.
Machine learning on process and inspection data — the discipline every fab hires for, taught with real datasets rather than toy ones.
Energy, water and abatement modelling for Indian conditions, feeding the Green Fab City model and the WA-07 and WA-08 curriculum.
Building teaching-grade instruments at a price a polytechnic can afford. Every one that works becomes a Fab On Wheels station.
Spares, subsystems and maintenance capability that India currently imports — mapped, prototyped and documented with partner MSMEs.
What actually improves outcomes in a hands-on discipline, measured against cohorts rather than assumed.
Curriculum, safety material and datasets published where we can, so the work outlives any single cohort or programme.
