I build things that require a computer and I mean that without restriction. Design, code, systems, infrastructure, the workflow that holds it all together. The discipline changes depending on what the problem needs. The thinking doesn't.
Right now: a structured multi-agent AI development factory — 30+ specialized agents with distinct domain expertise, persistent cross-session memory, an adversarial Evaluator/Builder loop, and an AutoOptimize agent that runs autonomous metric-improvement experiments. TypeScript runtime, works with any LLM including local Ollama.
Before that: ML notebooks (digit recognition, regression, Twitter OSINT scraping), Node.js chatbots, full-stack TypeScript apps. At some point the infrastructure became more interesting than what I was building with it.
Stack
TypeScript · Node.js · React / Next.js · Python · LLM / agent tooling · scikit-learn
Find me
sohampod.github.io · LinkedIn