Om K

I’m a Frontend Developer with internship experience building reusable React components and integrating REST APIs. At Hymech NDT Equipments, I improved UI responsiveness across multiple device resolutions and strengthened UI/UX consistency with backend teams.

I also build LLM systems that prioritize reliability and evaluation. In InsightForgeAI, I engineered an autonomous RAG + ReAct agent (LlamaIndex, GPT-4o, ChromaDB) that answers natural-language questions over PDFs with full page citations and zero hallucinations.

To keep quality measurable, I built a FastAPI backend with async agent execution and a React 18 + Vite frontend using optimistic UI, custom hooks, and a single-process Render.com deployment. I evaluated the pipeline with a Ragas harness across 20 ground-truth Q&A pairs, scoring Faithfulness, Relevancy, Context Recall, and Answer Correctness using automated HTML reports.

Beyond RAG, I develop explainable and security-aware AI. In ExplainableAI Intrusion Detection System, I used the NSL-KDD dataset to optimize a supervised Random Forest model for high recall and minimized false negatives, then added probabilistic risk scoring and transparent feature-level insights for actionable recommendations. I’m equally interested in agentic automation, demonstrated in my Agentic AI SDLC Automation Tool (CrewAI, FastAPI, React/Vite), where I generated complete SDLC artifacts with a 3-agent prompt-chaining pipeline and human-in-the-loop feedback.