I build and ship production-grade AI systems — from RAG pipelines and MCP servers to full Kubernetes MLOps pipelines. I don't just train models; I deploy them, monitor them, and keep them running.
Autonomous multi-step blog agent with MCP server integration, RAG pipeline for grounded content, web search, image sourcing, and SEO optimization. Full LLMOps: MLflow prompt versioning, FastAPI streaming, CI/CD on every push.
Complete ML lifecycle: data ingestion → training → evaluation → K8s deployment. Rolling deployments for zero downtime. GitHub Actions CI/CD gates on 85%+ accuracy before any deployment. MLflow tracks every experiment.
Production sentiment API with FastAPI async endpoints, Docker + K8s deployment. Class balancing improved minority class recall by 22%. Sub-200ms latency with full CI/CD pipeline — broken code never reaches production.