Designed and built a data streaming platform that powers the event-driven architecture of ChimeCore — Chime’s proprietary payment processor and ledger. The platform, built on top of Kinesis, provides strong correctness guarantees required for handling financial data, and supports features such as transactional publishing for DynamoDB and PostgreSQL, dead-letter queues, event replays, and more.
Led multiple cross-organizational initiatives to reduce cloud costs at Chime, saving $15 million annually. These efforts included database storage optimizations — archiving historical data to S3, compressing and de-duplicating datasets — as well as leveraging right-sizing and autoscaling strategies across various services.
Created a centralized repository for Chime’s Protobuf schemas, along with tooling and infrastructure for schema validation and multi-language code generation. Partnered with engineering teams across Chime to drive adoption and ensure consistent usage of Protobuf.
Technologies used: Kinesis, DynamoDB, PostgreSQL, Lambda, SNS, SQS, Airflow, Athena, Spark, Snowflake, Kubernetes, Terraform, Datadog, Go, Python, TypeScript, Protobuf.