- Completed and productionized the event stream integration initially designed by a Principal engineer.
- Configured Kafka, KafkaConnect, and Debezium to capture WBS changes and ingest them into a Budget-specific table.
- Built logic to conditionally execute queries against the new store.
- Monitored response times with DataDog and SumoLogic dashboards.
- Documented the implementation extensively in Confluence, enabling other teams to adopt similar patterns.
- Cut Budget report response times by 50–100% (from ~5s to 1–2s).
- Proved out a domain-specific data store pattern, influencing other Financials teams to use event streaming.
- Strengthened Procore’s broader shift toward microservices and service isolation.
- Implemented batching to handle high-volume event floods caused by nested WBS schemas.
- Reduced duplication by consolidating nested events, enabling stable Kafka throughput.
- Applied the Outbox pattern to safely stream changes between domains.
- Delivered a simplified, report-ready WBS schema purpose-built for Budget.
The first time I turned on the WBS event stream in production, it nearly overwhelmed the system. We underestimated how many nested events each WBS change could trigger. In real time, I had to design and implement batching and deduplication strategies to stabilize ingestion. Once resolved, seeing the Budget report drop from 5 seconds to under 2 was an incredible moment — and it marked one of my first experiences delivering a Staff-level systems solution.