What I led
Owned architecture and implementation end-to-end on AWS, including roadmap/timeline definition, indexing approach evaluation, performance metric evaluation, and cross-team coordination for migration and rollout.
Enterprise GenAI Retrieval Platform
Built in a financial-services context to migrate clients from a legacy index to OpenSearch, support vectorized document retrieval at multi-tenant scale, and improve front-end search performance for frequent daily usage.
Stack
AWS OpenSearchPer-tenant multi-index strategyAWS API Gateway + Lambda service layerIngestion Lambdas, search Lambdas, and embedding LambdasAWS Bedrock Titan Embeddings V2Amazon S3+4 more
Highlights
- Designed and implemented a multi-tenant retrieval architecture on AWS for hundreds of client tenants.
- Led migration planning and execution from a legacy index to OpenSearch.
- Defined and implemented document vectorization strategy (chunking + embedding generation) for new document ingestion.
- Evaluated indexing options across open-source and OpenSearch-aligned approaches, including Lambda-based patterns, then selected implementation direction.
- Implemented API-driven search flow where user queries are embedded and matched against OpenSearch vector indexes.
- Added intent-aware query expansion/reformulation and RAG-driven retrieval patterns to improve context-driven responses.
- Added analytics support around search usage and performance.
Outcomes
- Scaled multi-tenant retrieval workflows across 100+ client contexts and high document volumes.
- Supported thousands of user searches per day.
- Search latency improved materially versus the legacy indexing approach.