On-Prem REST API connector.
Performance, reliability, and real-time infrastructure for the On-Prem REST API connector team in BMC Helix’s product suite.
Container image reduced from 835 MB to 113 MB.
Illustrative workflow · explore the engineering above
The problem
Cross-platform binary compatibility issues were blocking deployments, and the connector image was 835 MB. Polling delayed the event flow. A race condition in Redis queue consumption caused duplicate request delivery when multiple containers ran concurrently.
What I built
Re-engineered the Docker pipeline as a multi-stage build and resolved the binary compatibility problems. Designed and implemented SSE across three microservices, backed by Redis request queuing. Tracked down and fixed the concurrent queue-consumption race condition. Authored container naming, profile management, and CLI tooling for multi-container orchestration.
The engineering thinking
The work covered both deployment and runtime behavior: reducing the image footprint, fixing compatibility, replacing polling, and addressing the race condition at queue consumption. The orchestration design made those improvements repeatable through the team’s setup scripts.
The outcome
Cut image size by 86%, from 835 MB to 113 MB, and unblocked cross-platform deployments. Replaced polling with real-time events and eliminated the duplicate-delivery race condition. The orchestration design became the team’s standard deployment workflow. Recognized among top performers and received a pre-placement offer.