Dipak Bundheliya

I’m an AI/ML Engineer specializing in GenAI and agentic automation, focused on building RAG-based pipelines, multi-agent workflows, and LLM-powered applications across compliance, sales, and computer vision. In my current role, I design an internal RAG compliance assistant that helps product teams conversationally query new US regulations and safety standards.

I’ve architected semantic vector search over compliance knowledge bases to retrieve relevant regulatory clauses and map them to product label requirements, while keeping document ingestion, chunking, and embedding strategies production-ready as standards change. I also build LLM chat interfaces for regulatory updates and compliance summaries—bridging research-grade ideas to tools teams can actually use.

Previously, as a Junior Data Scientist, I automated ML workflows for regression, time series forecasting, and anomaly detection—reaching 87% accuracy on dynamic datasets—and improved data processing efficiency. I’ve also delivered practical AI projects like a sales agent for lead scraping and outreach automation, multi-agent travel planning via Crew AI, and a shipping label detection system using YOLOv8 with OCR integration, all reinforced by an ownership-first mindset from idea to end-to-end delivery.