In a recent position at a new startup, I designed a mobile REST API in Node.js and Restify that proved to be performant, reliable, and easy-to-maintain. The codebase is distinctively separated into individual services and the requests are processed based on the router model.

The application is bundled into a Docker container and deployed onto multiple AWS EC2 instances via Docker Cloud in a continuous delivery fashion.

Performance and uptime were outstanding which was a great validation of the architecture.

A real-time tweet monitoring service that allowed users to track and archive tweets containing individual keywords was having trouble keeping up with volume after being acquired by Hootsuite.

I rewrote the keyword matching algorithm to mimic the keyword filtering behavior of Twitter’s streaming API, which allowed for reverse matching with user-defined keywords. The new algorithm outperformed the original one by several orders of magnitude thanks to achieving constant-time lookups on its larger data set.

While implementing these improvements, I also added a number of self-healing mechanisms that detect and reestablish degraded connections, resulting in very high uptime.