Forge

Computing Has Earned a Teaching Hospital

Our students build, and they build well. They ship working software in their courses, in Orbital, in hackathons, in open-source repositories that strangers depend on. The talent is here and the appetite is here. What a field this strong is ready for next is the one thing medicine takes for granted: a place to form its builders on real work, end to end, the way a teaching hospital forms a doctor on real cases. Computing has produced its specialists and its researchers. It is ready to build that ward.

SoC can be the school that builds it, and the next stage of a field this strong is to give its builders a place to be formed on real work, end to end, over time. That place is what I am proposing. A center the School runs, where students take a real organization's problem and carry it all the way to something used.

The Builder Who Sees the Work Through

The person this forms is rarer than a strong engineer, and the years ahead will ask for many of them. She sits with an organization that knows its problem but not the technology, understands what it actually needs, and finds the specification it never thought to ask for. Then she builds it, and stays with it until it is in real use, because the center that ships a thing also maintains it - so the work is not handed off at the demo and forgotten.

The industry has a name for the closest version of this person: the forward-deployed engineer, who works inside the customer's world rather than at arm's length from it. Forge forms that instinct from the start. The point is not that the graduate leads every change she touches. It is narrower and more honest than that. She understands the organization, builds the fit, and sees the work through.

The Formation That Belongs in the School

This formation belongs inside the School, because it reaches something a placement elsewhere cannot. An internship hands a student a defined task inside a defined team, for a season, at one company, and a good one teaches a great deal. But an early-stage startup, where many ambitious students choose to go, is a singular situation by its nature: the lessons learned there bind to that one company's particular mess at that one moment, and much of it does not transfer. That is the nature of the stage, not a failing of the host.

Forge gives the student the other thing. She owns the problem end to end rather than a slice of it. The stakes are real, because a real organization is waiting on the result. The mentorship is continuous, not seasonal. And across a body of engagements she sees many organizations, not one. The depth comes from ownership held over time.

Earned by Building, Open to All Who Build

Every SoC student may walk up to the door. None walks through it on a promise. Entry is a building test: ship working software and land a real contribution to something people already use. The artifact decides, not the pitch.

For a long time the right way to back a young builder was to hear them out and grant them the means, because building was expensive and a pitch was the only evidence available before the fact. That model was sound for its time, and it produced a great deal. What has changed is the cost: with today's tools a student can build the thing itself in the time it once took to describe it. So the fairer filter is now the better one. We can ask to see the work, not the promise of it. The door is open to everyone, and the bar is the same for everyone, whether they arrive through an open application, a course such as IS3251, or CIP. Open at the door, high at the bar.

Formation That Feeds the Faculty's Own Work

A teaching hospital does not station a senior doctor over a junior one to catch mistakes before they reach the patient. The senior doctor is there to form the junior into a doctor. The patient is protected, and the trainee is made. Both at once. That is the duty Forge takes on, and it is a duty owed in two directions.

When students sit with an SME, a faculty member and a working practitioner sit with them. The faculty member brings the discipline of why a thing works; the practitioner brings the scars of building it under real constraints. The student gets both, on a problem that has a business depending on the answer. The SME gets people who are accountable for the result, not advice from a distance.

There is something in this for the faculty too. The problems an SME actually carries in are rarely the tidy ones in a paper. They are the open questions. A supervisor who helps a student solve one is holding live research material, drawn from the field rather than imagined in a seminar room. The teaching feeds the work, and the work feeds the teacher.

By the Time They Lead, Everyone Knows Why

Students advance in Forge the way capability is recognized everywhere it is taken seriously: by what they have built, and by who relied on it.

A student starts on a defined piece of a real engagement, supervised closely. They ship it. The client uses it. A maintainer merges their work into something that runs. Each of these is a small fact about the world that was not true before, and each one is checkable. As those facts accumulate, the student is handed more: a wider slice of the problem, then a whole engagement, then the responsibility of helping lead one and bringing newer students through it.

The stages are real. A student starting out, a student who has proven they can carry a build, a student trusted to lead one. Each is earned by the body of work behind it, not granted by the calendar. Responsibility grows to meet proven capability, and the proof is public. By the time a student leads, everyone around them already knows why.

The Hands That Complete the National Effort

Singapore has put real weight behind AI, and it shows. There is a national plan above S$1 billion, a generation of training and enablement that AI Singapore has built into genuine capability across the workforce, and now OpenAI's first Applied AI Lab outside the United States, opened in Singapore alongside a forward-deployed-engineer bootcamp around 20 May 2026. Awareness, funding, tools, training. The ground has been prepared.

For the largest firms, this is enough; they have the engineers to build on it. For smaller firms it is not yet enough, and the figures say where the gap sits. In 2024, AI adoption among SMEs ran at 14.5 percent, against 62.5 percent for larger firms. The plan and the tools have reached them. What turns a tool into a working system inside a particular business is the people who can build it.

What completes it is implementation: a pair of hands that understands the firm and builds the fit. A computing school is the natural place to produce them.

The Standing Source on What Works for SMEs

Every engagement Forge takes on starts as a real question nobody has answered yet: what does this firm actually need, and what will it take to make it work where they are. Answered enough times, across enough different firms, those questions stop being one-off jobs. The patterns that hold across many firms - what works in practice and what does not - become a body of knowledge Forge can share openly, with each client's own code and specifics kept confidential.

This is, in miniature, what a frontier applied lab does. It works on real deployments, learns what generalizes, and publishes. The difference is who it is pointed at. A frontier lab's economics send it toward the largest contracts and the most valuable customers; the small firm with a stubborn, specific problem will never clear that bar. Forge is built to serve exactly those firms, because students power the work.

The claim here is modest and worth making carefully. Not a research institute. A growing, practical record of the de-identified patterns - what works when real technology meets real small businesses - kept in the open for the next builder to use, while each client's own build stays its own.

More for the Whole School, Honestly Costed

A new center invites a fair question: does it draw from a pool the rest of the School depends on? Forge is built to enlarge that pool rather than draw it down. The engagements throw off a steady supply of real problems and the data that comes with them, the kind that frontier labs guard and most faculty cannot easily reach. They open standing relationships with industry. They give the newest faculty something especially scarce: live problems, student builders, and grant work with adoption already attached. They raise the School's standing as the place where this is done. And Forge earns its own income, so the resources it brings in are additional, not redirected.

The honest cost sits inside that gain, and it is not recovered for some time. In the early years Forge will take real faculty supervision hours and return nothing measurable for them - no paper, no grant, only a student more capable than before. That is the bill, and it falls on the people whose time is already the most contested in the School.

It Funds Itself, and Fits the School

A dedicated academic lead of practice-professor standing runs Forge: not a duty bolted onto an already full load, but the work itself. The engagements are paid at the market rate, so the center covers its student stipends and its lean operating costs from the value it creates. At steady state it asks the School for no recurring budget.

An educator weighing this would ask the real questions in one breath: who owns the software a student builds for a paying client, what happens when a build fails, whether students are protected rather than worked cheap, and how a center serving outside clients sits inside a university at all. None of these is new, and each has a bounded answer. The client owns what is built, and the School keeps a teaching and research license to learn from it. Engagements are scoped, insured, and run within the structures the University already uses for outside work, so a build that fails is a contained engagement, not an exposure. Students are paid for the work and formed by it, on bounded hours, never cheap labor. Settling those terms precisely is the first real work once the idea has backing, and none of it is ground the University has not crossed before.

It will, honestly, need some startup help to stand up - administrative support through the first cohort, while the engagement pipeline and the billing find their feet. That is a real early cost, and the center carries the burden of growing out of it.

Beyond covering itself, Forge could in time return a surplus toward other innovation work across SoC. That is a direction to grow into, not a number to promise today.

A Lead Your School Can Take, at Low Risk

Across roughly forty of the world's leading university programs, every individual element of Forge is already running somewhere and working. Open admission on a hard bar; entry by built work; clinical-style supervision; real client engagements; student-run operations. What no program has done is combine them into one center that builds, deploys, owns what it ships, and is co-run by the students who pass through it. Each part is proven, but the combination is not: making the pieces hold together as one working center is the real, unproven work, and some of it will only be learned by doing it. That is also the opening, and being first to put the whole together is the lead your school can take.

The closest model in Singapore is the SMU-UOB Asian Enterprise Institute, which pairs students with external industry mentors on real, project-based consulting for SMEs - genuine, well run, and a credit to the idea. Forge goes one step further: its students build and deploy and maintain what they make, and gradually come to run the center themselves. The full landscape, with named precedents, sits in the annex.

The Proving Ground for What CIP Forms

Forge stands alongside the C-Innovators Program, not beneath it. CIP forms the leader the years ahead will need - breadth, judgment, the habit of carrying an ambition across technology, business, and society. Forge is where that capability is forged on real work, and where it becomes visible: a student who has understood a real organization, built what it needed, and seen it through to use has shown what the formation produced, in public, on the record.

The two fit because they answer different halves of the same question. One forms the person. The other proves what the person can do.

What this needs from you now is a decision only you can make, and a reversible one: back a single first engagement, and name the person to run it. Not a budget, not a building, not a standing commitment - one cohort, one client, watched closely, and stopped if it does not earn its keep. The supervision it will take is real, and I have not hidden its cost. But the case for the next step is not made on this page. It is made the first time a student sits with a real business, builds what it needs, and sees it through, in public, on the record. Your school is ready to be the first to build this. The first engagement is where it starts.

What Other Schools Do - the Landscape

I surveyed roughly forty leading universities. Every element of Forge already exists somewhere; the table below shows where, and where each precedent stops short.

Before committing to Forge, I looked at 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 this region, in both their computing and their business schools. The closest precedents to Forge are real and instructive - and each is scoped or selected in a way Forge is not.

Program	Where	What it does	Where it stops short of Forge
Real-client clinics
Data Science for Social Good	UChicago → CMU	Small teams + full-time mentors build for real government / non-profit partners	Scoped to data-for-good, not general SME software
Data Science Discovery	UC Berkeley	Credit-bearing, always-on clinic at scale (~250 projects and 900+ students a year)	Same data-only scope; not a general build clinic
Hack4Impact · Cal Blueprint	Penn · Berkeley	Students ship real, deployed software for non-profits ("deploy, not demo")	Non-profit-scoped; student-org, not a center with supervision + progression
Consulting on real SME projects (the closest local model)
Asian Enterprise Institute · SME Consulting Program	SMU-UOB	Undergraduate teams take on real SME projects under external industry mentors (subsidized)	Students advise (reports, recommendations), they do not build or deploy; project-based, and they do not run the center
Entry test & verifiable record
Google Summer of Code · Outreachy	Open-source	Stipended real contributions to live open-source projects; selection is doing the work	Not a university; no SME client work
Creative Destruction Lab	U of Toronto	Holds a high bar throughout - culls between every session	Selective at entry; venture-focused, not open
Apprenticeship & work-integrated
Mayfield Fellows	Stanford	Paid, embedded startup apprenticeship (12 students/year)	Tiny and selective; sends students into others' startups
Overseas Colleges (NOC)	NUS	Paid startup internships abroad, for credit - Singapore's flagship learn-by-working model	Sends students out to others' startups; Forge keeps them in, owning real client work
Furnace	NUS SoC	In-school incubator (deferred-fee) - the in-school precedent	An incubator for ventures, not a teaching hospital doing supervised client work

The closest local comparison is the SMU-UOB Asian Enterprise Institute's SME Consulting Program - undergraduate teams taking on real SME projects under external industry mentors. It is a good model, and it marks exactly where Forge goes further. There, students advise: they study a business and hand over recommendations. In Forge they build: they ship and deploy working software, and own whether it actually works in the business. Their work is guided by industry mentors alone; Forge blends practitioners with faculty, whose involvement also turns each engagement into material for their own research. And there, students are project-based visitors; in Forge, the senior ones help run the center itself.

Reading the full dataset, four positions are conspicuously empty - and they are exactly the ones Forge takes:

• Open access held to a high bar. Programs are open-and-easy or selective-at-entry; almost none stay open to all yet hold a hard bar throughout via continuous public proof.
• Entry by a building test, not a pitch. No university I could find gates entry on both a deployed build and a merged open-source contribution.
• A general software clinic for real businesses. The clinic model exists only for data-for-good or non-profit software, never as a general teaching hospital for SMEs.
• Resourcing judgment, not capital. Even the 2024-26 AI wave still resources building (credits, GPUs, funds); almost nothing resources judgment, experience, and proof.

Full methodology and the institution-by-institution dataset are available on request (working files: the landscape survey of ~40 institutions across computing and business schools, US / UK / Europe / Israel / Greater China / Korea / Japan / Singapore).