At Zvolv, I built production RAG pipelines for Zai using Python and FastAPI. They handle document ingestion, embeddings, vector search, and grounded responses, with tenant-scoped isolation.
I improved retrieval quality through chunking strategies, context assembly, and retrieval optimization, helping improve grounding and reduce hallucinations in customer-facing AI features.
For Zinie, I built LangGraph agent workflows with tool calling, enterprise data access, and multi-step orchestration. I also implemented failure handling and recovery when downstream services fail.
At Zvolv, I led the migration of legacy PHP functionality to Node.js and TypeScript across a platform serving 25+ enterprise customers. Earlier, at RainClouds Global Solutions, I developed REST APIs and implemented authentication and access controls for a cloud-based security platform; on a personal project, I integrated MCP with Notion so an AI agent could answer natural-language queries about business records.

