I build backend systems that stay fast when the load doesn't let up.

Architected an agentic validator that audits e-applications for defects and auto-generates fix-sheets, turning a slow manual review loop into an automated one and cutting issuance time by 72.3%.

Engineered a reactive orchestration engine on Java Easy Flows and MongoDB Change Streams, using virtual threads and a reactive WebClient to bring end-to-end latency from 20s down to under a second.

Built a Scan-Push service on Kafka with a factory architecture to stream shipment updates to clients — handling 2B+ updates daily via multi-threading and AsyncIO. Also led the FastAPI + Kafka integration layer for a TMS SaaS MVP that onboarded 6 pilot clients.

Transformed the Waybill service with multi-threading and Vert.x for a 10× drop in event-processing time and fault tolerance worth a 3× business gain. Separately overhauled billing with a Spring/Spark rewrite and Drools caching in Redis, cutting execution time by 70%.

Added end-to-end observability to an agentic-AI application with self-hosted LangFuse, surfacing the prompt and workflow refinements that reduced token usage — the kind of visibility that turns an LLM app from a black box into something you can tune.

Architected the RAG engine — LangChain, LangGraph and QdrantDB — that reverse-engineers ATS scoring, and set up the CI/CD (GitHub Actions + Docker) that automates releases. Now one of the most-starred resume tools on GitHub.