Samay Jain

I’m an IIT Roorkee B.Tech graduate with experience in AI/ML engineering, agentic AI systems, RAG pipelines, and LLM application development. I focus on building production-oriented Generative AI solutions that automate knowledge workflows while prioritizing safety with guardrails and structured execution.

As an AI Intern at Helic AI, I designed an automated DevOps knowledge acquisition pipeline using the Stack Overflow API plus Selenium and BeautifulSoup, turning community QA into a retrieval-ready dataset and cutting manual collection effort by 70%. I also built a LangChain-based multimodal RAG pipeline for PDFs, images, and screenshots using vision/OCR-assisted extraction and LLM summarization. During my Time Series Research Internship at MNIT Jaipur, I analyzed 100+ organizations with Granger Causality and ML to quantify causal links between employee sentiment and stock performance (statistically significant p < 0.05) and identified sentiment clusters tied to measurable stock shifts.

In my projects, I’ve delivered end-to-end agentic RAG systems: an Enterprise Kubernetes documentation intelligence pipeline with LangGraph/LangChain, FastAPI, and NeMo Guardrails to reduce unsafe query pass-through by 80%. I implemented an LLM gateway with automatic failover to improve uptime (99.9%), plus observability and evaluation to reduce hallucination risk by 20%. I’ve also built a multi-agent dental appointment assistant with planner/executor/validator guardrails and persistent SQLite memory, and VisionTalk—a YouTube video Q&A RAG system using hybrid retrieval (Dense + BM25 + reranking) for 40%+ answer relevance gains.