██████╗ ██████╗ █████╗ ████████╗██╗██╗ ██╗
██╔══██╗██╔══██╗██╔══██╗╚══██╔══╝██║██║ ██╔╝
██████╔╝██████╔╝███████║ ██║ ██║█████╔╝
██╔═══╝ ██╔══██╗██╔══██║ ██║ ██║██╔═██╗
██║ ██║ ██║██║ ██║ ██║ ██║██║ ██╗
╚═╝ ╚═╝ ╚═╝╚═╝ ╚═╝ ╚═╝ ╚═╝╚═╝ ╚═╝
AI Engineer & DevRel · Galileo
Interested in high performance agentic systems
I've spent the last several years across the AI stack: evaluating agents without ground truth, building continual improvement systems for transformers, and making search systems that actually work in production. I've shipped these as products, written about them in 4 books, and open-sourced the benchmarks so others can build on them.
Before Galileo, I led full-stack AI at Enterpret and Morningstar, going from zero to production each time.
|
Agent Leaderboard Open framework for benchmarking LLM-based enterprise agents. Hallucination Index Featured in Andrew Ng's newsletter. BRAG Family of open-source small LMs fine-tuned for RAG. Matches SoTA at under $25/model to train. |
|
Founded Maxpool in 2019, a community of professional AI engineers, researchers, and builders. Focused on agents, evaluation, RAG, and the business of AI. Also built Max, a custom AI assistant integrated into the community. |
🥇 1st place in every internal hackathon at Galileo, Morningstar, and Enterpret 📈 Top 20 nationally in Microsoft's Search Ranking challenge 📝 Top 20 in SEMEVAL |
Latent Space interview · Talks at international communities · Top AI Developers to Watch 2023
Best way to reach me → x.com/ptkbhv · linkedin.com/in/bhavsarpratik