AI systems that hold up
once they're in production.
I'm Chetan — an AI/ML engineer who builds RAG pipelines and multi-agent systems, then instruments them with evals and tracing so their behavior is measured, not assumed. Currently building risk-intelligence infrastructure at Hansraj Ventures.
A few systems worth a closer look
From production risk-intelligence platforms to open-source retrieval systems — each one paired with the evals to prove it works.
Risk Minor
ProductionA multi-agent risk-intelligence platform: hybrid BM25 + Milvus retrieval feeding a RAG pipeline, with RAGAS and Langfuse wired in for real-time hallucination monitoring.
FamioX Suite
ProductionTeams-style real-time collaboration backend — WebSocket channels, presence, and threaded messaging serving concurrent sessions at sub-100ms delivery.
Fact Check
Open-sourceMulti-agent fake-news detection — three local LLMs queried in parallel, verdicts aggregated via a consensus layer to suppress single-model bias.