iPic

A mentoring role. The product was written by a student team. My part was the architecture,the process and the review sessions — not the commits.

The brief

A student team had three months to ship their end-of-course project: a Pinterest-style photo sharing app. The hard part was never the feature list. Three months is usually enough to write the code and not enough to turn that code into something that runs on somebody else's machine — which is exactly where student projects die.

How I scoped it

We fixed the stack on day one and did not revisit it: Angular on the front, NestJS for the API, Firebase for auth and image storage, Docker so everyone ran the same thing, GCP to deploy on. Not because it was the optimal set, but because every choice had an obvious way back and documentation a beginner could follow.

The condition I set: deploy before you finish. Get something live on GCP early, however empty, and the rest of the term is just filling it in — rather than praying the night before the deadline.

What I coached

• Module boundaries and layering, agreed before the first line of code.
• Docker for the dev environment, so "works on my machine" stopped being a sentence anyone said.
• A CI/CD path to GCP, stood up early and pressed by the team themselves.
• Weekly code reading: I pointed at what was wrong and why, they did the fixing.

How it landed

The team defended on time and iPic was voted best project of the course. The part I rate higher than the award: by the end of term they were deploying without asking me.