Kumail Raza

I’m an Applied AI Engineer with 6+ years in software, focused on shipping agentic, LLM-powered products in production. I build AI systems on AWS Bedrock (Claude) using LangGraph and AG2/AutoGen—multi-agent orchestration, planner-based routing, tool calling, and RAG with vector search—grounded in real business data.

I also extend agent capabilities with MCP (Model Context Protocol) and focus on LLM evaluation and tracing with LangSmith. From token streaming to Bedrock Guardrails, I design systems to be safe, observable, and reliable in real user workflows.

My full-stack foundation helps me own the whole path from model and architecture through to a deployed, observable product. I work with Next.js, React, and Python FastAPI, and I manage multi-account AWS infrastructure using CDK—ECS Fargate, RDS, Cognito, and more—so the solution stays cohesive from backend to user experience.

In my role as a Senior Applied AI Engineer, I design and ship agentic AI systems on AWS Bedrock with LangGraph and AG2/AutoGen, deploying multi-tenant platforms in daily production use. I’ve delivered end-to-end AI products with enterprise SSO across identity providers, automated testing and disciplined PostgreSQL migrations, plus hardened security (AES-256-GCM secret management) and CloudWatch/X-Ray observability. Earlier work includes building full-stack and frontend systems, and projects like CLEA, Arcade, and enterprise agent platforms built for production performance and cost control.