At Godash.ai, I built a 25-case evaluation suite for an LLM support chatbot, checking intent routing and page selection, with an LLM judge to flag unsupported claims.
I also developed its vectorless RAG agent as a LangGraph state machine, with structured routing, cited answers, per-thread memory, and streaming from FastAPI to a React widget. I logged conversation latency, token usage, routes, and errors in PostgreSQL.
At SigniMinds, I built an AI agent stack for a real estate site, including an OpenAPI spec, an A2A Agent Card, and a WebMCP manifest. Its AI-readiness score rose from 45 to 93/100, and it passed 77/77 checks on a second benchmark.
At the AI Research Center, Woxsen University, I built a RAG course assistant with ChromaDB vector search that served 5,000+ students at 89% retrieval accuracy. I’m a final-year B.Tech student focused on quality engineering for AI applications, and I also built a Dockerized YouTube ads compliance auditing service using LangGraph and Vertex AI Search.

