Harsh Chainani

I’m an AI/ML-focused engineer with a background in Artificial Intelligence and Data Science, and I love turning research ideas into working systems. I’ve built agentic and ML pipelines that solve concrete problems through retrieval, detection, and robust inference.

My projects include a Retrieval-Augmented Generation QA agent using LangChain and vector databases, exposed as an MCP tool for LLM-client querying. I also built a U-Net lung CT segmentation model and evaluated robustness against FGSM and PGD adversarial attacks. For security, I developed an autonomous cyber defense agent using anomaly detection (Isolation Forest and decision trees) with simulated SOC workflows and automated response logic.

I’ve contributed to applied work through a multilingual AI chat & voice assistant (10+ languages) and an AI job-matching pipeline using LangChain/LangGraph with structured extraction. I secured 2nd place in an AI/ML hackathon for the voice assistant, and I co-authored research on proactive threat mitigation published at ICSICE 2026.