Careers · Engineering

Can you still teach yourself software engineering?

AI has made the first working version of anything cheap to produce. Everything after that first version is still engineering, and that part has not moved.

Vin Lim Founder, Astralab
Download

Sin Chew Daily, the Malaysian Chinese-language newspaper, ran a careers feature on our founder Vin Lim in December 2025. The profile covers the career itself. What seemed worth writing down is what the interview says about how people actually learn to build software, and which parts of that AI has changed.

Can you still teach yourself software engineering?

Yes, and the raw material has never been easier to reach. That is worth stating plainly because the honest version of the story is less romantic than it sounds. Vin learned from books in secondary school, because the web then had far less on it. He took a multimedia diploma and could not afford to keep studying after it. What followed was a job, then a company that folded when the pandemic stopped the events business it depended on, and then Astralab in 2020.

Someone starting now has the opposite problem. There is so much material that the difficulty has moved to knowing what to skip. The constraint was never access. It was the number of hours between reading something and being able to build with it, and no amount of available material shortens that.

What does AI change about learning to build software?

It moves the floor, not the ceiling.

Someone with an idea and no engineering background can now describe what they want and get something that runs. That is a real change and it is not a small one. But the thing you get back has small faults distributed through it, and finding and fixing those is the job. Past a certain point the work stops being about producing code and starts being about the shape of the system: what happens under load, what happens when a call fails, where the data actually lives, which decisions are expensive to reverse later.

That is the part a prompt does not do for you. Vin's own position is that models still hallucinate, and that code produced this way is not reasoning about the larger structure it is being added to. We build with these tools daily and we still read what they produce, for the same reason.

So the answer to whether AI removes the need to learn this properly is that it removes the need to learn it in order to start. It does nothing about the part where you have to finish.

Where does the learning actually happen?

Mostly around other people doing the same work, which in Penang in 2013 was hard to arrange: the developer community was small and had little local support. So Vin started GDG George Town, the Google Developer Group for the city. It has run at least two community conferences a year since, bringing engineers from outside Malaysia to speak to a local audience. Most events are free.

The useful observation buried in that is about incentives. Running the group is not charity; organising a conference forces you to find people who know more than you do and then spend two days listening to them. It is a reasonable way to keep learning after the point where structured courses run out.

The Sin Chew Daily careers page of 18 December 2025: a full-page Chinese-language feature on how Vin Lim taught himself into a career in digital technology, with photographs of him speaking at and attending developer events
星洲日報, print edition, 18 December 2025. Reproduced with credit to Sin Chew Daily; reported by 黄梓清.
Sources

「職業指南針」 career feature, Sin Chew Daily (星洲日報), print edition of 18 December 2025; online edition published 17 December 2025. Reported by 黄梓清; photographs supplied by the interviewee. The print page is reproduced above and downloadable at the top of this post, with credit to Sin Chew Daily. The article is in Chinese; the piece above is our own writing, not a translation.