Software Engineer @ Procol
Jun 2023 – Dec 2024
Owned backend performance and platform features across analytics, sourcing, and intake products.
The problem
Analytics dashboards and the three main sourcing APIs were too slow for real-time procurement decisions. At the same time, every new client integration was a bespoke engineering project that delayed go-live.
The approach
- Profiled hot paths to find where time actually went instead of guessing.
- Refactored slow queries, added targeted Redis caching, and removed redundant data fetches for analytics.
- Tuned the three highest-traffic sourcing APIs end-to-end.
- Built Integration Studio as a configurable adapter-and-mapping layer so new platforms slot in without per-client code.
Impact
Analytics response time cut by 40%
Sourcing API latency reduced by 80%
Integration Studio led to 3× more integrated clients
What I learned
- The bottleneck is rarely where you first think it is.
- A small number of queries usually drive most of the load.
- Productizing internal tooling is often the highest-leverage thing a platform team can do.
Tech
Ruby on RailsPostgreSQLRedisAWSPerformance Optimization