At Optimizely, I led delivery of the Experimentation Remote MCP Server, prototyping and building six production tools that let AI agents manage experiment entities. I also proposed the integration architecture adopted across the Opal platform and drove readiness, reliability, and correctness for its experimentation tooling.
I replaced Optimizely’s production documentation RAG system with Google Vertex AI Search, redesigning the retrieval pipeline to improve answer accuracy and reliability. Earlier, I optimized the Statistical Significance service, cutting costs by 98%; outside work, I build open-source AI systems including Askable and Converse.
