At Julia AI, I own an AI-powered code search platform serving 1,200+ daily users across four products. I built its multi-agent orchestration, production RAG pipeline, and evaluation frameworks, with the RAG pipeline achieving 2.1-second P95 latency.
I established evaluation methods adopted company-wide, achieving a 22% MRR lift, and built a complexity-aware LLM router that reduced inference costs by 60%. I also implemented responsible AI guardrails and human-in-the-loop approval workflows for production agents.
At Hitachi Vantara, I optimized Java and Spring Boot APIs, led an eight-service monolith decomposition, and designed an event-driven pipeline processing 2M+ events per hour. My project work includes multi-agent systems, responsible AI code review, Llama fine-tuning and serving, and distributed task processing.
