- 🏥 1.5+ years in AI/ML applied to clinical trials — real-world ML, not just toy projects
- 🔌 Building & contributing to LLM gateway / agent infrastructure (Rust + Python SDKs)
- 🤖 I build automation agents that research, scrape, and answer — running 24/7 from a phone
- 🎯 Open to GenAI engineering roles (Pune / Maharashtra / Remote) and freelance LLM-integration work
- 🧘 Side quest: a Sanskrit-epics RAG bot that answers questions about the Gita, Vedas, Ramayana & Mahabharata
- Running ASWP — an open, preregistered-style experiment: does training-only weight noise, annealed to zero, beat a deterministic baseline in Transformer training? Pre-results protocol, engineering spec, SAP + adversarial design review
- Open-source contributions to UmaiTech/aura-llm-gateway — a high-performance Rust LLM gateway implementing the Open Responses API:
- Dogfooding my own Agent-Reach fork as my daily research tool
- Shipping a Reddit Top Posts Digest Apify actor (top posts from any subreddits, no API key, RSS fallback) — open-sourced on GitHub
- Launching painfinder — mine Reddit for your niche's pain points (ranked themes, verbatim quotes, sources). Free & MIT, with live demo
- Shipping cutshort-cli — search Cutshort jobs and read full JD's from the terminal, no API key, zero third-party dependencies (pure stdlib). Live page
- Writing the agent-era learning guide — a 30-day roadmap for what to learn when AI writes the code. "Generation is commoditized; verification is the new moat."
What I actually do with them: LLM gateways & proxies · RAG pipelines · vector search · agent tooling · data pipelines · scraping & automation · API integration · MCP
- Rust (LLM gateway internals — routing, cost tracking, streaming)
- Open Responses API spec & agentic workflows
- Paid open-source contribution strategies (bounties, mentorships)
- GitHub: kragent66-glitch
- Open to: GenAI engineering roles · freelance LLM/agent work · open-source collaborations
Building the plumbing that makes AI agents work — one PR at a time.



