Still becoming something interesting.
I don't have a 10-year plan. I follow interesting problems and occasionally get obsessed with them. This is the commit graph.
Major versions run down the main branch. Detours, projects, papers, and open questions live where they actually happened.
Born in Pune. No roadmap, no architecture docs, just a lot of curiosity and a surprisingly durable attention span.
Early habit of saving strange questions before knowing what any of them were good for.
Learned to code. Realized software could be a lever for almost anything, then started pulling every lever within reach.
Small utilities and web experiments built because some local annoyance got loud enough.
Tried too many stacks too quickly. Useful for taste, less useful for finishing things.
Experiments
Half-finished clones
local-onlyA graveyard of almost-products that taught scope control the slow way.
Went deep on ML and systems. Published. Failed a lot. Learned more from the failures.
Datasets, annotation, retrieval, evaluation, and the weird places language models become overconfident.
Trying to understand why great engineers see the same problem differently before they can explain why.
Shipping products, thinking about systems at scale, still asking too many questions.
Tools that reduce context reconstruction and make teams faster without pretending humans are optional.
Why some ideas spread while better ones die. Not just quality, not just truth, something about carrier fit.
In progress. The next major version is still compiling, which is either exciting or suspicious.
Open to problems that would justify a major version bump.