Forge

A Teaching Hospital for the School of Computing

Singapore's medical students learn medicine in a teaching hospital, treating real patients under supervision from their earliest years. Its law students staff legal clinics. Its computing students build a great deal – in their modules, their capstone projects, and their internships – but almost always for a grade, or as a junior contributor inside someone else's company. What they rarely do is own a real problem for a real organization from end to end, answerable for whether it is genuinely adopted and used. That is what neither an assignment nor an internship is built to teach.

Forge closes that gap. It is the School's own teaching hospital: a centre students go to, earning their way in by building, then solving real problems for real businesses under supervision, with real consequences. The students who come through it are Smiths – deeply technical builders who can take a real organization's problem, shape a working solution, and lead the people around it. The market has begun calling this builder a forward deployed engineer; a Smith is that and more, built not only to deploy into an organization but to lead it. Not engineers narrowly executing a spec. Not consultants advising from the outside. Smiths: the ones who make the thing that works.

Forge is open to every student in the School. There is no application essay and no pitch. The way in is a building test, because a school that builds should admit students by building, not by talking. And the bar stays high throughout: students advance by what they ship and who adopts it, judged by reality rather than by a panel. Open to all, demanding of all.

Forge stands alongside the Computing Innovators Programme, not underneath it. CIP forms the kind of leader the years ahead require; Forge is where that capability is forged on real work and made visible. The relationship is set out in full further down – it is the part that makes both programmes stronger.

A New Kind of Builder, and the Moment That Demands One

AI is no longer a function added alongside the others. It is becoming the substrate on which an organization runs. Work that used to require whole departments now runs on systems, and a company is increasingly designed by its leaders, not merely managed by them. The scarce person of the decade ahead is the one who can architect that design and lead the humans at its core – technically deep enough to build it, broad enough to lead it. That person is neither today's engineer nor today's executive. The pipelines that produce each were built for an era when they were different people doing different work. That era is ending, and no major university has yet redesigned its programme to produce the new kind on purpose.

At the same time, the cost of building has collapsed. For two decades, the way a university supported student innovation was to select a promising few, give them a grant, a desk, and a mentor, and hope a company emerged. That was a rational answer to its constraints: building was slow and expensive, capital was scarce, space and guidance were hard to find. Every one of those constraints has dissolved. A student with a cloud subscription and an AI assistant can ship in a weekend what once took a funded team a quarter. The grant, the desk, and the gatekeeping committee now answer questions the world has stopped asking.

When building becomes cheap, the scarce things change. What is rare now is not the ability to build. It is the judgment to know what is worth building, the experience of solving a real problem for someone who actually has it, and the proof that you can do the work. None of these can be delivered in a lecture hall or bought with a grant. They are forged only by doing real work for real people.

Medicine has the teaching hospital. Law has the clinic. Architecture has the studio. Each is an institution where students do supervised real work as the core of their training, not an afterthought to it. Computing has no equivalent – and so its graduates, capable as they are, have rarely owned real external work from end to end before they leave. Forge is that missing institution, and it is newly possible precisely because cheap building has, for the first time, put real contribution within a student's reach.

An Unoccupied Position

Before committing to this, we surveyed how the world's leading universities promote innovation and entrepreneurship – roughly forty institutions across the United States, the United Kingdom and Europe, Israel, Greater China, Korea and Japan, and our own region, in both their computing and their business schools. The finding is clear, and it is the heart of the case.

Every individual mechanism Forge uses exists somewhere. Real-client work appears in data clinics and student build organizations. Staged ladders are near-universal. Equity-free positioning is common. The AI-native framing is now crowded. But no institution combines more than two or three of Forge's elements, and none combines them all. Forge's originality is not any single feature. It is the synthesis – and several of the positions it occupies are, across all forty institutions, simply empty:

Nothing here is an untested bet. Every piece already works somewhere; no one has assembled them into a single programme. NUS can be the one that does – one of the strongest computing schools in Asia, in an economy dense enough to put new talent to work the moment it is ready.

The Work the Nation Actually Needs Done

Singapore has bet, at the national level, on becoming an AI-first economy. The hardest unmet part of that bet is not the frontier labs or the large enterprises – it is the small and medium businesses that form the backbone of the economy and that, for the most part, have no realistic path onto AI. They lack the talent, the budget for consultants, and the time. Getting SMEs onto AI is precisely the kind of problem that is easy to declare and very hard to do.

The gap is not a hunch; it is measured, and it is widening. By IMDA's own Digital Economy Report, AI adoption among large firms in Singapore reached 62.5% in 2024 – against just 14.5% among SMEs. A year earlier the two figures were 44% and 4.2%: SME adoption more than tripled in a single year, and the distance to the large firms still grew. This is despite more than a billion dollars committed under the National AI Strategy and a dense stack of support programmes.

And it is not, at root, a digitalization gap – 95.1% of SMEs have already adopted some digital technology. It is a building gap. Among the SMEs that do use AI, 84% rely on off-the-shelf tools, against 44% of large firms. The large firms cross the line because they can build systems that fit how they actually work; the small ones cannot, so they are left with generic packages that mostly do not.

This is why the national effort has not closed it. The support that reaches SMEs comes in three forms – grants that make AI cheaper, catalogues of pre-approved tools, and advisory services. Each is real and useful, and not one of them is a pair of hands that builds. A subsidy does not write the integration; a pre-approved catalogue is the very reason SMEs stay on off-the-shelf tools; advice produces a recommendation, not working software.

National programme	Who it reaches	Form of support
For SMEs
SMEs Go Digital · GenAI Navigator	SMEs	Grants (up to 50%) + pre-approved tools
CTO-as-a-Service · GenAI Sandbox	SMEs	Advisory + trial environments
Productivity Solutions Grant	SMEs	Subsidy for pre-scoped solutions
For large firms
Enterprise Compute Initiative (S$150m) · GenAI Playbook	Large / mature firms	Compute + bespoke partnerships
Champions of AI	Selected leading firms	Capability support
MAS FSTI 3.0 (S$100m)	Financial institutions	AI / quantum capability grants

This is the work Forge's students do. An SME's real problem – a process that should be automated, a system that should be rebuilt, an AI capability it cannot reach alone – is exactly a problem of redesigning how an organization works. That is the Smith's craft, practised on real organizations rather than rehearsed in a classroom. Every engagement does double duty: a student is forged on real business experience, and a Singapore business moves a step further onto AI.

The positioning writes itself: NUS computing students as the people who carry Singapore's businesses onto AI. It is good for the students, who graduate with real experience and a record to show for it; good for the businesses, who get working software; and good for the School, whose students become visibly central to a national priority.

There is a name the industry has started using for this person: the forward deployed engineer – the technical builder a company sends in to sit with a problem, work out what is really needed, and build the fix on the spot. It is among the most sought-after and hardest-to-fill roles of the moment, and no degree sets out to produce it. Forge produces it as a by-product: deploying a Smith to an SME is forward-deployed work by another name, and a Smith leaves ready to do it anywhere. The value also runs back inward. A forward deployed engineer carries what they learn in the field home to the organization behind them, and here that organization is the next cohort: each engagement returns as knowledge the newer Smiths inherit rather than start without, so every problem solved in the field makes the whole programme better at the next one.

And this is not a bet against the grain – it is the same bet the market is now making, one tier up. In May 2026, OpenAI chose Singapore for its first applied AI lab outside the United States: a centre built to deploy AI against real-world problems, opening a Forward-Deployed Engineer bootcamp to train builders for exactly that work. Forge is the student-scale version of the same idea. Where a frontier lab deploys senior engineers into large institutions, Forge forges the next generation inside the School, on real problems for the smaller businesses a global lab will never reach. When the world's leading AI lab plants its first overseas applied lab and an FDE bootcamp in the same city, the signal is hard to miss: this is the capability the moment is asking for, and a School of Computing can produce it at the source.

Forge and the Computing Innovators Programme

Forge and CIP serve the same person by different means, and that is what lets them reinforce each other without either being folded into the other. CIP forms the leader. Across four years it builds the disposition, the breadth, the cohort, the aspiration, and the Showcase – the whole-person formation that turns a capable technical student into someone who can lead an organization. Forge is where that leader does the work. It is the open arena, on real business problems, where capability becomes visible contribution. CIP is the formation; Forge is the proving ground.

Because Forge is open to everyone, it does something CIP could never do for itself: it makes CIP's value measurable. The non-CIP students in Forge are the baseline – what a capable computing student does on real work without the formation. The gap between that baseline and how CIP students perform – how fast they clear the bar, how far they advance, the quality of the feedback their clients give – is the visible evidence of the difference CIP makes. A formation programme can rarely prove its worth; Forge is the instrument that can. (Read rigorously, CIP students are already selected for ability, so a fair comparison controls for where they entered rather than crediting the whole gap to formation.)

And this is why Forge's openness protects CIP rather than diluting it. The two draw their prestige from different sources: Forge's from the hardness of a bar anyone may attempt, CIP's from a selective, multi-year formation and the leadership trajectory it opens. Forge widens the floor and surfaces talent; CIP selects and forms the few. A strong record in Forge becomes one of the clearest signals for CIP admission – so an open Forge feeds CIP a deeper, already-proven pool, while CIP remains the thing that turns proven builders into leaders. Neither sits beneath the other.

How It Works

Forge runs on a simple loop. Students earn entry by building. They take on real problems from real businesses. They do the work under supervision. They advance by what they ship and who adopts it. Each stage is a bar set by reality, not by a committee – which is how a programme stays open to everyone and demanding of everyone at once.

The Gateway: No Pitches, Pull Requests

The primary way into Forge is a credit-bearing gateway course. This is deliberate: it gives the entry a real home in the curriculum, funds the teaching that prepares students, and makes "open but selective by effort and competency" literally true – anyone may enrol, and you earn your place by what you build, not by what you claim. The course assesses the two faces of real software work:

The course is the main door, but not the only one. A student who can already clear the bar may enter through an open building challenge run off-cycle, or by an existing track record of shipped and merged work. The door is wide; the bar is real. And the gateway assumes students use AI – that is the modern craft, not cheating. Problems are set hard enough that the tools alone will not clear them, and a short live session, in which a student extends their own work and defends a design decision, confirms the work is genuinely theirs.

Real Problems, Real Businesses

The work is not simulated. A continuous stream of real problems flows in from small and medium businesses with genuine needs and no easy way to meet them. Students claim a problem and build the solution, and the bar for success is the most honest one there is: the business actually adopts and uses what was built. This is where students learn business – not from a case study, but from sitting across the table from someone whose problem is real, working out what they actually need, and being accountable for whether the software works. This is not an internship. In an internship a student is a junior pair of hands in someone else's company; in Forge a student owns a real problem end to end, under teaching supervision, and the point is not the labour but the learning – the work is the curriculum.

Supervision

A teaching hospital does not send first-year students to operate alone. Every engagement runs under supervision: the more senior Smiths who have cleared the higher bars mentor the newer ones, and faculty and experienced practitioners review work before it reaches a real business and step in when the stakes require it. Supervision protects the businesses Forge serves, and it is where the deepest learning happens – in the review, the correction, and the standard held.

A Workshop, Not a Row of Desks

Forge is a shared workshop, not a set of solo assignments. Smiths work alongside one another, and the hardest-won knowledge – how to actually get an AI solution working inside a real business, what holds up in deployment and what quietly fails – is pooled rather than re-learned one student at a time. In that sense the work is also research: each engagement is an experiment in deploying AI for a real organization, and what it yields becomes a growing, practical body of knowledge the whole programme draws on. It is, in miniature, what a frontier applied lab does – the difference being that here the people learning to deploy are students, and the learning is the point.

How Smiths Advance

There is no graduation ceremony and no panel that promotes you. Smiths advance the way every craft has always advanced – by a growing body of work that speaks for itself: problems solved, software shipped, contributions merged, businesses served. That record is public and verifiable, and it becomes the most valuable thing a student carries out of the School.

The door is open at the bottom and the bar is real all the way up. Nobody is selected in; everybody is pushed against a standard the real world sets. Most students will not reach the top, and that is the point. The bar is the teacher.

The Operating Model

This is the part that sets Forge apart from how universities usually fund this work. Student entrepreneurship is normally paid for by grants and the school's budget – a cost the institution carries, year after year. Forge funds it from the value it creates. It is designed to pay for itself – which makes it, fittingly, the proof of its own thesis: an entrepreneurship centre that is itself entrepreneurially self-sustaining. The mechanics are straightforward, and the details below are starting points to settle together, not conditions to resolve before the idea can move.

Forge is run as a centre of the School of Computing – a recognized, accountable home of the kind the School already hosts for mission-driven work (in the spirit of institutes such as the Asian Institute of Digital Finance). That gives it a place to sit, a clear line of governance, and institutional standing, without asking the School for new money. The finer points of structure are a conversation to have together; what matters here is that the model is sound, self-funding by design, and modest to begin.

What It Takes to Start

Forge needs almost none of what innovation programmes usually ask for. No grant pool to give away, no co-working space to build, no large new budget – because the resources that used to be scarce are no longer the constraint. What it needs is a mandate and a small amount of operational support.

None of this is a large financial commitment, and none of it is a recurring one: Forge is built to sustain itself, and in time to help sustain the entrepreneurship work around it rather than draw from the same budget. What it requires is a decision: that the School of Computing should train its students the way every serious profession trains its own – on real problems, for real people, held to a real standard – and that this is worth doing now, while the position is still open and the national need is acute.

What I want for them

I have spent years teaching computing students, and too often the same thing happens: gifted young builders graduate having built plenty in their courses, yet rarely having owned a real problem for someone who genuinely needed the answer – and they get hired as engineers when they could have been so much more. My ambition for them is larger than a good first job. I want them to leave able to walk into any organization, see what it really needs, build it, and lead the people around it – the rare person who turns a problem into a working thing and carries others with them. I want them to have done real work for real people long before they graduate, and to leave with the judgment and the proof that only real stakes can give. That is the builder Singapore's next decade needs, and the one I have spent my career trying to develop. Forge is how I would do it at scale – and I am asking for the mandate to build it, and a single semester to prove it with a first cohort.