Kumar Kartikay
Software Engineer at Google, exploring ML research in my free time.
I'm a software engineer at Google in Bangalore, working on large scale ML-powered data pipelines that help us improve Search Quality. Outside work I dig into machine learning research, most recently at the Lossfunk residency (batch 8), where I explored tree-search-guided LLM reasoning.
I like building things end to end, from distributed log pipelines to small reinforcement-learning engines, and I keep a few of them running in the wild.
Wrapping up the Lossfunk residency this week, building tree-search methods for LLM reasoning.
A personal journal, logging, and task-management system I built for myself. I use it every day!
A visualization of Monte Carlo Tree Search applied to LLMs.
A compact C++ AlphaZero implementation, agressively optimized for a home PC with a single GPU, using async pipelines and boost fibers.
Research and learning notes, coming soon.