Smallstep Labs, a digital identity startup, hired me to build a subscription billing solution. I proposed and developed a decoupled Stripe implementation in Python that exposes a GRPC API to Smallstep’s microservices. In follow-up projects I helped set up data reporting using Zeppelin and Terraform. I also laid the dev-ops groundwork for building Cloud Dataflow jobs.

Twitter's growth meant the job of scaling never ended, and this was certainly true for revenue processing. A cool solution I fondly remember introducing, was the concept of double sharding. We spread write load by uniformly distributing data across database clusters and then structurally sharding that data within the cluster for process optimized reading. This helped us deal with the onslaught of chargeable event data resulting from the introduction of Twitter's video ads.

In addition to safe guarding the revenue processing system, I focussed on simplifying tooling and processes with the goal of reducing friction for large advertisers and agencies.