Hi, I'm Krithik.
Robots by day. Search engines and inference deep-dives by night.
The bet underneath all of it: AI amplifies what you actually know. So I build the fundamentals from scratch first, use LLMs hard to learn faster, then point them at problems with the clarity that only comes from understanding the layer underneath.
A few things about me
- Shipped a deep-dive on LLM inference — Llama 3.1 8B on an H100, end to end. Wanted to understand what actually happens between hitting enter and the tokens showing up.
- Writing a search engine in Rust from scratch. Started chapter-by-chapter through Manning's IR textbook, now running BM25F over AWS docs as a real corpus. 3 new pieces
- Built filmsearch — semantic search over 54K films. v1 failed because I was throwing LLMs at the problem. v2 worked because I built BM25 with multi-zone ranking myself.
- Built a Claude skill grounded in Skycak's Advice on Upskilling. It refuses to let me move on from a concept until I can retrieve it cold without help. The reason the inference article and the search engine work exist — both required learning a domain from zero in a few weeks.
- Matic Robots — building home robots. On the assembly line: crown modules, functional tests, defect triage.
Past
- PM at Saayam — consumer mobile. Shipped chat coordination, cut task-completion time ~20%.
- PM intern at Wurq — fitness tech. Owned the ML pipeline for pose correction. 490ms → 67ms inference, edge costs ~28% lower.
- Ops at ZS Associates — pharma consulting. Owned optimization on a platform handling $32M in rep payouts. Killed ~35% of support volume.
- M.S. Engineering Management, Northeastern · B.Tech. CS, VNIT Nagpur.
Around the web
Looking for applied-AI roles close to customers — APM, forward-deployed, technical product. The kind where you're building the thing and shaping how it goes to market at the same time. Inference, search, agentic systems, dev tools — anywhere the work is real and the layer beneath the LLM still matters.
If you want to talk — building something, hiring, or just curious — easiest paths below.
Piqued your interest?
→ the full story — how I got here, what I'm working on, how I think about LLMs.