Hi, I'm Swathi

I'm going through a renaissance moment building AI things with AI.

My current focus:

  • Making William Gibson's quote about the future being unevenly distributed less true. I work with domain specialists and companies to apply AI meaningfully into their practice.
  • Studying model quirks by using a text-only prompt to create a game across 30+ model-effort combinations

Past life:

  • I've built or distributed products at two Y-Combinator fintechs and ByteDance (Lark). Each company was digitally transforming how businesses in Southeast Asia worked or got paid.
  • I graduated with an LLB at NUS Law (QS Asia #1) and was called to the Bar in Singapore. But I found creating things in tech more fulfilling.
  • At 14, I came to Singapore on a science scholarship seeking a challenge and it's been home ever since.

Fun fact: GitHub preserved code I wrote for an ML data-science project in an Arctic Vault in 2020.

Nothing I share here should be considered legal advice.

Latest Tinkering Swish I use the same prompt across different model-effort pairings (eg: astra max, fable 5.1 ultracode, opus 5.5 high) to build a 2-level, 3D game. I measure what it takes to make the game (time, token, cost), and whether the game was worth making (immersiveness, playability, fun). + more
Latest Thinkering Diagram over a painting of a woman resting in a lavender field: human intent plus AI capability drives execution, abstraction, expression and domain renaissance, which expand the frontier of creation, so more people can build, new kinds of things can be built, and builds begin building themselves; more ideas become attemptable, which raises ambition and feeds back into human intent The more we can do, the more we do, the more there is to do Why you can never do enough with AI + more