Building AI systems that retrieve, reason, call tools, and ship as real products.
divijworks.com · LinkedIn · X · Email
I work on practical GenAI systems: RAG pipelines, agentic workflows, MCP servers, tool-calling agents, evaluation loops, and full-stack AI interfaces.
Unstructured data -> retrieval -> reasoning -> tool calls -> workflow -> product surfaceThe AI layer should not be a thin chatbot wrapper. I prefer systems with a working loop:
- Retrieval with citations, reranking, memory, and source traceability
- Agents that call real tools and expose their state clearly
- Interfaces that let users inspect, correct, approve, and trust outputs
- Backend pipelines that can be deployed, observed, and improved
- Evaluation paths that make behavior measurable instead of vague
- Production-grade RAG and agent systems
- MCP servers for real operational tools
- AI products with inspection, approval, and feedback loops
- Voice/screen AI and live context systems
- Applied AI for incidents, compliance, legal workflows, and developer tools
Open to serious AI engineering work, GenAI product collaborations, MCP/RAG systems, and full-stack AI builds.


