Iâve touched on this before, but it feels more urgent now.
The way we think about work in software engineering has fundamentally shifted. Back in the day, whether it was app dev, data engineering, or applied ML, the model was fairly straightforward: write first, build later. Managers managed, engineers coded, and the cycle repeated.
But with GenAI, the landscape is not the same. Tools like Cursor and agent frameworks have slashed entry barriers. The sheer lucrativeness of âjust codingâ is no longer what it used to be. Managers, too, are strugglingâbecause the field itself has moved to build-first, iterate-fast.
So where does that leave us? I see two clear paths:
Deep Research & Core Tech
Work on the fundamentalsâmodel training, fine-tuning, RL, quantisation, system-level innovation. This is where specialized labs are emerging, and many talented folks are already moving into these spaces.
High-Velocity Product Building
Donât just codeâship. Use AI as leverage to move faster than ever. Build-first, validate quickly, and keep pushing the cycle.
Both paths are demanding, but both are rewarding. The middle groundâjust being a âcoderâ in a world where AI codesâwill shrink fast.
To developers in Sri Lanka: rethink your trajectory. Aim to either deepen your craft into research or raise your velocity in product building. The future rewards those who choose boldly.
The way we think about work in software engineering has fundamentally shifted. Back in the day, whether it was app dev, data engineering, or applied ML, the model was fairly straightforward: write first, build later. Managers managed, engineers coded, and the cycle repeated.
But with GenAI, the landscape is not the same. Tools like Cursor and agent frameworks have slashed entry barriers. The sheer lucrativeness of âjust codingâ is no longer what it used to be. Managers, too, are strugglingâbecause the field itself has moved to build-first, iterate-fast.
So where does that leave us? I see two clear paths:
Deep Research & Core TechWork on the fundamentalsâmodel training, fine-tuning, RL, quantisation, system-level innovation. This is where specialized labs are emerging, and many talented folks are already moving into these spaces.
High-Velocity Product BuildingDonât just codeâship. Use AI as leverage to move faster than ever. Build-first, validate quickly, and keep pushing the cycle.
Both paths are demanding, but both are rewarding. The middle groundâjust being a âcoderâ in a world where AI codesâwill shrink fast.
To developers in Sri Lanka: rethink your trajectory. Aim to either deepen your craft into research or raise your velocity in product building. The future rewards those who choose boldly.
