What people say after completing the tracks

I left school twenty years ago and the idea of writing code felt completely out of reach. The Roots exercises were harder than I expected in the second and third week — the statistics section especially — but the written feedback from the tutor was specific enough that I knew exactly what to re-read. By week seven I was producing notebooks I could actually explain to someone else.

Track: Roots

I had tried two other deep learning courses before Branches — both were video-only and I stalled after a few weeks because I had no way to check whether my understanding was correct. The four projects here were genuinely demanding. The code review on project three picked up a design decision I'd been making without realising it, which explained why my validation loss kept behaving oddly. That kind of feedback is hard to find.

Track: Branches

The Canopy programme is long, and by month four I was genuinely tired. But the fortnightly check-ins with my mentor kept the architecture decisions grounded in something real. He pushed back on my first data pipeline design in a way that saved me from a fundamental problem I wouldn't have caught until much later. Having a hosted project page at the end — something I can actually link to — made the length feel justified.

Track: Canopy

The office hours twice a week were the most useful part for me. I'm working full-time and I can't always articulate what I'm stuck on until I've sat down and tried to explain it to someone. The tutor was patient and didn't move past a question until I could restate it in my own words. I finished Roots feeling like I understood the material, not just that I'd completed it.

Track: Roots

Eleven hours a week is a serious commitment on top of a full-time role. I'll be honest — weeks eight and nine were rough and I considered deferring. The GPU access was something I hadn't expected to matter as much as it did: being able to run a proper training job instead of a reduced demo changes how you think about the whole process. I'd do it again.

Track: Branches

My background is in data engineering so I came in knowing pipelines but not model design. The Canopy mentor understood that distinction immediately and calibrated the review sessions accordingly — more time on architecture and less on infrastructure. The technical report defence was more demanding than I expected, but the writing workshops in the final weeks helped considerably.

Track: Canopy