At Faire, I architected a multi-tenant Generative AI platform on AWS Bedrock, integrating Claude, Mistral, and Titan behind a unified inference gateway. I also designed a RAG system for merchant product discovery using OpenSearch vector search and hybrid retrieval.
I directed ML and cloud engineers through GenAI initiatives and established MLOps standards with SageMaker Pipelines and Terraform. I designed event-driven services for asynchronous LLM inference and optimized embedding and vector indexing pipelines on Pinecone and OpenSearch.
At Pinecone, I designed vector search infrastructure for enterprise RAG applications and built LLM inference optimization pipelines using Hugging Face Transformers and FAISS. At Databricks, I supported distributed ML training pipelines on Spark MLlib and contributed to MLOps tooling for experiment tracking and model versioning.
