The Best Way to Start a Tech Career in the Post-AI World

One idea that matters more than the whole list

Whether you are a student choosing what to study, already in tech and thinking of pivoting, or trying to break in, you do not want to spend three years getting good at something AI automates or the market abandons. So before the list, one idea: tech jobs do not usually disappear. They change.

Think of people who managed physical servers and networks. Cloud computing came along and automated or moved much of that work. The smart ones did not sit around complaining. They learned AWS, Azure, DevOps, cloud architecture and security. They moved into the thing replacing the old way. The same is happening with AI.

So do not ask "will AI replace my job?". Ask: which part of my job will AI replace, and what does a human become responsible for after that? That gives you far better answers.

AI and machine learning: highest ceiling, highest bar

If you genuinely enjoy maths, statistics and programming, this has one of the highest ceilings in tech. I mean people who understand models, data, training and evaluation, not someone who made a few API calls to OpenAI last weekend.

I disagree with the idea that you always need a PhD. To be a frontier researcher at OpenAI, Anthropic or DeepMind, you are competing with the strongest academics in the world, and a PhD probably matters. But plenty of applied ML roles just need someone who understands the technology and can solve real problems with it. If you like maths and code, this is an excellent path. If you hate maths and are choosing it because AI is hot, do not. For the distinction, see AI vs machine learning.

AI engineering: software engineering with AI

This is probably more relevant to most readers. An AI engineer takes models built by Anthropic, OpenAI or Google and turns them into something useful: an AI tutor, an assistant, an internal agent that searches documents and performs tasks, a support system. They are not training models from scratch.

That is why I like the role: it is basically software engineering with AI. You still need APIs, databases, deployment, security and authentication. On top, you learn how models behave, how agents work, retrieval, context and evals. Here is my prediction: AI engineer will not stay a special job title forever. Nobody calls themselves an "internet software engineer", because the internet just became part of software. AI is heading the same way, and it will become table stakes for every engineer. I wrote a full roadmap in how I'd become an AI engineer from zero.

Software engineering: still my top pick, learned differently

I am biased, because this is where I came from. But yes, I still strongly recommend software engineering. The only thing I would change is how I learned it. Five or six years ago, you spent a ridiculous amount of time memorising syntax and writing every loop by hand. You still need to understand programming, but that is no longer where most of your value comes from.

Value moves upwards. Can you understand what the business needs? Talk to a customer? Turn a vague idea into concrete requirements? Design a system, use AI to build it, test the output and ship it? The data shows this: traditional computer programmer roles are projected to decline while software developer roles are projected to grow. A programmer who receives instructions and types code is easy to automate. An engineer responsible for an outcome is not. One good engineer gets far more productive with AI, and expectations rise to match.

Cybersecurity, cloud and DevOps

Cybersecurity will stay extremely valuable. I would not call it AI-proof, since AI will automate parts of it, but demand has an obvious driver: more software, more cloud, more employees using AI tools, more people shipping apps they barely understand. Every new system or connection is another point of attack, and AI makes attackers better as well as defenders. It is a huge field: penetration testing, application security, cloud security, incident response, governance and compliance. If you naturally look at a system and think "how could someone break this?", you will probably love it. The caveat: it is not an easy escape from software engineering. Many people are trying to move in, so you still need real skills. See our cybersecurity careers in Malaysia guide.

DevOps, cloud and platform engineering are not glamorous. Nobody grows up dreaming of infrastructure. But everything has to run somewhere. You can build the coolest AI product in the world and someone still has to deploy it properly, scale it, secure it, keep the database alive, monitor it, and stop it collapsing when traffic spikes. Nobody cares about infrastructure until it breaks, then everyone panics. If you like understanding how things work under the surface, this is a very solid career. Just do not make your identity "the AWS guy". Tools come and go; the underlying problems stay.

Product management and tech sales

Product management becomes more important as software gets cheaper. If building is ten times cheaper, companies could build ten times more, but should they? Usually not. The bottleneck becomes what is worth building: what customers actually want, which problem we are solving, which idea to prioritise, which feature sounds great internally but nobody will use. The hard part is communication, judgement, prioritisation and getting engineering, design, sales and marketing to agree. If you love five quiet hours with headphones on one problem, product may drive you mad: your calendar will be meetings. If you like people, ambiguity and defending a clear rationale for path A over path B, it is a great fit.

Tech sales, especially sales engineering and the newer go-to-market (GTM) engineer, is one of the most underrated paths. Many people like technology but not coding and assume tech is not for them. Wrong. Some of the most valuable people in tech companies can understand a complicated product and explain it so a customer gets it. Nobody spends millions on cybersecurity, cloud or enterprise AI because a salesperson said "trust me". They want someone who can sit with their engineers, answer hard questions and run a real demo. If you are technical, commercially minded, good with people and motivated by money, this might be your path, and because you are close to revenue, the upside from commission can be very high. Read tech jobs that don't require coding for more options.

Industry matters almost as much as job title

Two software engineers. One maintains internal systems at a small traditional company. The other works at a high-growth AI company where software is the product. Same title, completely different trajectory: different learning curve, salary growth and upside. If money matters to you, do not just ask what job to do. Ask where the money is flowing. Right now: AI, fintech, cybersecurity, cloud infrastructure and high-growth software.

And if you are in Malaysia, your market does not have to be Malaysia. If your skills are globally competitive, companies in Singapore, Australia, Europe and the US are options. Malaysians generally have good English, and combined with a technical mind that can take you a long way. The economics of your career change completely. Our remote tech jobs guide covers how.

What I would actually do if I started from scratch

For most people, I would go the software engineering path, but learn AI from day one. I would learn how software works: databases, APIs, backend systems, deployment and the fundamentals. At the same time I would use AI actively, build AI features, understand agents and work with LLMs. In short, become a software engineer who is extremely proficient with today's AI stack.

Then specialise. Love maths? Go deeper into machine learning. Love infrastructure? Cloud. Obsessed with security? Cybersecurity. Good with customers and business? Product or tech sales. And if you are, say, an accountant, those same skills let you become the accountant who does the work of ten, which changes your salary too.

You do not need to predict which job will be hot in 2035 if your fundamentals are strong today, and software engineering fundamentals are hard to go wrong with. Stop hunting for an AI-proof career; it probably does not exist. The safer strategy is becoming someone who adapts whenever the technology shifts, which is what has happened in tech again and again. That fundamentals-plus-AI foundation is exactly what our 12-week AI-Native Software Development Programme builds, and you can try the first lessons in the free trial. Not sure which path suits you yet? Try our career compass.