Aditya S.

I build AI agent systems that ship to production. Not prototypes, not demos. Production systems handling real users and real data. At work - I contributed across three major AI features from scratch: - An AI search engine that went from 30% to 70% accuracy through system design alone. Query classification, multi-step planning, dynamic attribute resolution. No fine-tuning, just better architecture. - A Slack AI agent shipped across Python backend, Go API gateway, and Vue 3 frontend. OAuth flows, access-controlled scoping, Block Kit formatting. Launched to 136 connected customers. - An MCP Apps ecosystem: 28K lines, 30 tools, 200 tests, cross-platform compatibility with ChatGPT, Claude, and Cursor. I also built GSTflo, a multi-agent tax OS using LangGraph (6 autonomous agents, full audit automation). And ContentEngine, an end-to-end AI video pipeline that goes from script to published YouTube Short with zero manual steps. My stack: Python, TypeScript, Go, FastAPI, Next.js, LangGraph, LangChain, MCP servers, Docker, Kubernetes, AWS. I've also written C++ trading engines and CUDA neural nets when the job called for it. I work fast, communicate clearly, and deliver working systems. Not decks. Not proposals. Code.