At JIIT, I built CyberSentinel with a two-person team. Our five-agent intrusion detection pipeline won the Agentic AI for Cybersecurity track at JAI 2026.
I built the backend and agent orchestration, including fusion logic that reconciles agents’ verdicts into an explainable policy. CyberSentinel scored 96.25/100 on the hidden replay set.
On Trace, I built a LangGraph pipeline that turns hand-drawn architecture diagrams into scaffolded code using Gemini and ChromaDB. I also designed operation-level canvas diffs so the LLM patches the canvas instead of regenerating it.
I led a four-person team building TenderHawk, AI web agents that check government tenders against an MSME’s eligibility and cite the relevant bid clause. I built its agent layer and scraping pipeline, with validated bid data and live agent steps streamed to the dashboard.
