Ashish kumar

I’m a Senior Full Stack & AI Engineer with 5+ years building production AI/LLM systems, with the last 3 years focused on RAG, agentic workflows, multimodal extraction, and computer vision. I lead technical direction end-to-end, from architecture to delivery, and I actively mentor engineers to raise team velocity.

At my current role, I designed an agentic AI workflow (LangGraph + Gemini) that pulls domain data from multiple sources and delivers insights with human-in-the-loop controls—reducing manual research effort by 30%. I also built a real-time notification service using PostgreSQL WAL-based CDC, where an LLM classifies and prioritizes events and pushes alerts to 1,000+ users with ~20s latency over WebSockets.

I’ve delivered high-impact extraction and query experiences too: a Text-to-SQL engine using schema-grounded RAG (75% accuracy across 200+ schemas) and a multimodal pipeline pairing YOLOv11 layout detection with vision-LLM parsing for 90% field-extraction accuracy. I’ve also re-architected a legacy monolith into a scalable multi-tenant SaaS with secure SSO, improving throughput by 20% and cutting login time by 40%.