Navigating the AI Job Market: What Actually Matters
AI is moving fast, but the job market is not keeping up. If you are trying to break into AI roles right now, most of what you read online will mislead you.
This post is based on a talk by Andrew Ng and Laurence Moroney. It covers the current AI job market, what actually moves careers forward, and a few traps worth avoiding.
The market is tight, but not closed
The last couple of years saw a wave of overhiring. Companies pulled back in 2024-2025, and entry-level roles got harder to land.
But the opportunity is still there. Companies are just more careful now. They want people who can ship working software, not people who can describe AI at a high level.
Three things that actually move careers forward
Laurence Moroney breaks it down into three pillars.
Understand your domain deeply. Not just the surface level. Know the models, the trade-offs, and what is actually happening in the field versus what is just noise.
Align your output to business needs. Technical work that does not connect to a real business outcome is hard to defend. The more clearly you can tie your work to measurable impact, the better.
Bias toward delivery. This is the most underrated one. A lot of people can talk about AI. Far fewer can show working software that solves a real problem.
The bottleneck has shifted
AI tools are making code faster to write. But faster code just moves the problem. The new bottleneck is knowing what to build and being able to describe it clearly.
Product thinking is now a real advantage for engineers. Teams that used to run one PM for every 6 to 8 engineers are moving closer to a 1:1 ratio. If you can talk to users and write a clear spec, you stand out.
Not all AI is big AI
The industry is splitting in two. Big AI means large cloud-hosted models like GPT and Gemini, built for scale and general use. Small AI means lightweight models that run on-device, with lower latency and no data leaving the device.
Small AI is growing fast, especially in healthcare, law, and entertainment where data privacy matters. Phones from Vivo and OPPO are already running AI workloads on-device. On-device AI is a growing area of real work, not just a trend.
How to filter the noise
Social media rewards engagement, not accuracy. "Software engineering is dead" gets clicks. It is not true.
The skill worth building is learning what to take seriously versus what is just loud. A flashy demo is not a product. An AI agent that works sometimes is not a production solution. Stay skeptical of sweeping claims, and focus on what you can build and ship.
Technical debt in AI code
AI can write code fast. That is useful. But code you do not understand is a liability.
Relying on generated code without reviewing it creates systems that are hard to debug and harder to maintain. Think of it like financial debt. Good debt is an investment in something that grows. Bad debt is something you take on without thinking. The same applies to code you ship.
What this means practically
The people doing well in AI are not the ones with the most certifications or the most buzzwords in their bio. They ship working software consistently and stay close to what the business needs. They build strong networks and stay curious without chasing every new announcement.
The market is harder than it was two years ago. But it is not closed. The approach just has to be more deliberate.
Based on a talk by Andrew Ng and Laurence Moroney. Watch the original on YouTube.