At Apple, I build production generative AI systems for Apple Retail, including RAG pipelines, agentic workflows, and real-time inference. I led the design and deployment of retrieval systems that improved relevance by 22% and cut irrelevant results by 18%.
I engineered LangGraph agent workflows with function calling and MCP-based integrations, and built evaluation harnesses to measure hallucinations, factual consistency, and retrieval accuracy. I also adapted foundation models for retail classification and extraction with LoRA/QLoRA.
For AI features, I built Python REST APIs that sustained 8K+ requests per second at 99.95% uptime. I instrumented production systems for monitoring and optimized inference costs and latency across Claude models served through AWS Bedrock.
Earlier at Apple, I built AI platform infrastructure and Siri NLP analytics and data pipelines. That work included real-time telemetry ingestion, training datasets, and dashboards supporting evaluation and model development across search and voice assistant experiences.

