At Oracle, I designed and shipped the AI gateway and multi-agent orchestration services behind an enterprise assistant with 100K active users. I also built an agentic RAG workflow that cut average response latency from five seconds to two.
I raised document-based RAG answer accuracy from 56% to 87% by orchestrating topic-specific agents over enterprise knowledge articles. I built evaluation datasets from production sessions to catch routing regressions and agent failures before release.
To replace a paid model, I fine-tuned open-source Llama 3.2 3B Instruct weights, raising production routing accuracy from about 70% to 90% while reducing LLM inference cost. I also built REST APIs and MCP integrations, and supported Kubernetes deployments on OCI.
At FICO, I deployed 28 credit models for APAC and built a Kafka/Flink fraud-detection pipeline that increased throughput from approximately 300 to 500 records per second. I also developed a Python/LangGraph workflow to monitor production ML pipelines, and built a multilingual real-time voice assistant as a project.
