Vishwajitsinh Mahida

Python, LangGraph, TypeScript, React, Ollama, ClickHouse, Docker. Generalist with a focus on AI agent infrastructure and observability systems.

LLM evaluation methodologies, distribution-based anomaly detection for agent traces, and quality-regression alerting. Digging into ClickHouse for observability at scale.

AgentPulse — framework-agnostic observability for production AI agents. Just shipped a quality-warning evidence adapter based on 586 real traces.

Technical discussions on AI agent observability, LLM eval design, and metadata-only tracing. Always happy to talk about local LLM deployment or LangGraph patterns.