At Mobileum, I build production GenAI and backend systems, including multi-agent workflows, a database-agnostic RAG platform, and LLM tracing that captures reasoning, tool calls, MCP calls, latency, and token usage. I also own a human-feedback pipeline that turns user feedback into prompt improvements without fine-tuning.
I've delivered a real-time conversation-fraud model with 91% validation accuracy and roughly 90% recall, cut database infrastructure costs by 50%, and made API data processing 40x faster through connection pooling and session caching. My foundation spans Java, Python, Spring Boot, FastAPI, Kafka, Redis, PostgreSQL, and nine years of production-management experience in high-availability industrial environments.
