I was a founding engineer on the Financial Reports team, and owned this pipeline through two full scaling redesigns as it went from an initial build to a GA product serving real user query load.

• Built the v0 data pipeline as a founding member of the team: Airflow for orchestration, Kafka for ingestion, and Postgres materialized views to serve queries.
• As the first scaling step, split the pipeline into two independent services — a processor writing from the ingestion pipeline, and a reader serving user queries — separating write load from read load.
• After GA, full materialized-view refreshes stopped scaling with data volume. Redesigned the ingestion pipeline around daily, quarterly, and rolling aggregate tables in Postgres, computed asynchronously by a worker pool instead of recomputed on read via materialized-view refresh — designed and led the implementation, cutting API latency in half.
• Brought integration-test execution into the Kubernetes-based deployment pipeline.
• Introduced Architecture Decision Records to make technical trade-offs explicit and documented going forward.