DEVANSH SHUKLA

I’m a Full-Stack AI engineer focused on turning complex, real-world problems into reliable AI systems—especially with agentic approaches, LLM pipelines, and computer vision. I enjoy building from system design through deployment, so solutions are both effective and usable.

Recently, I’ve been developing AI-driven compliance models that automate regulatory risk assessment using RAG, advanced data chunking, and vectorization. I also build RAG pipelines that process complex legal documents with computer vision and edge-aware thinking, cutting manual audit effort significantly.

I’ve contributed as a Research Engineer on an enterprise Active Learning Platform, where I architected backend and AI integration layers and designed a microservices architecture for video transcription and context-aware LLM workflows. I’ve also led computer vision work on an assistive “smart blind stick” using YOLO-Tiny on Raspberry Pi, optimizing inference for low-power edge environments.

Across projects and hackathons, I consistently ground AI outputs with verification—like strict RAG pipelines for health recommendations and neuro-symbolic health tooling. I’m motivated by scalable AI architectures, educational impact, and building high-quality systems through open-source and strong engineering practices.