This is the project I live in day-to-day. It's a distributed product built on a microservices architecture — multiple services talking asynchronously, third-party integrations, automated workflow triggers. My job is to keep it healthy and move it forward.

The interesting work is in the gaps: debugging a backend issue that only shows up in production GCP logs, writing migration scripts to fix a database inconsistency without downtime, or translating a client's "it's broken" into a specific code fix. I've also been the person on the call with the client — which has made me a much more precise communicator.

The work here was translating raw database structures into something a non-technical stakeholder could actually act on. I wrote the SQL to find patterns and bottlenecks, then built the dashboards in Metabase and Tableau to make those findings visible.

One thread I'm proud of: SEO analysis through SQL — identifying which headers and load-time issues were affecting search performance, then making targeted optimisations based on that data rather than guesswork.

I designed and ran the SQL training program for five consecutive batches of new engineers — off-campus hires and fresh graduates. The curriculum was grounded in our actual production data, not contrived examples, which meant trainees could apply what they learned on day one of real work.

Teaching forced me to understand our systems at a deeper level than feature work alone ever would.

This was the first project I owned end-to-end. The brief was simple: a quiz platform for internal webinars with a live leaderboard. The reality was more interesting — real-time state synchronisation across many concurrent players, network drops mid-session, people joining games already in progress.

I built the core playing engine in Go-Fiber, used Redis to manage live session state, and PostgreSQL for persistence. The front-end was Nuxt / Vue 3. I also handled bulk quiz creation via CSV import, secure authentication, and Dockerised the whole thing so deployment was trivial.

The edge case work — handling browser reloads and reconnections gracefully — turned out to be the most technically satisfying part of the project.

My first production contribution. Apricot 3 was an internal tool to digitise the company's hiring pipeline — removing manual steps, reducing the time between a candidate applying and an offer going out.

I implemented Google SSO authentication, rebuilt core workflows to eliminate excessive page transitions, and built funnel analytics dashboards that tracked candidates across time, cost, and stage complexity. Built using Python — my first time working in a large, opinionated codebase. A good introduction to reading before writing.