Open-source ML homeworks with auto-tests - fundamental agorithms from first principles
This year I’ve been designing homework assignments for an ML course at Skoltech (Russia’s answer to MIT/Caltech for science and technology). After bombing more job interviews than I care to count, I think I’ve finally figured out what I was personally missing during my studies - a deep understanding of a relatively small set of fundamental algorithms. Well, my pain is the next generation’s gain!
In my engineering worldview, you can’t truly understand something unless you’ve built a replica from scratch with your own hands. At the same time, I didn’t want learning to stall at the terror of a blank page. I wanted to guide students toward each problem step by step. Show them how it’s assembled from small building blocks.
Once I’d settled on how to frame the problems, the remaining question was how to grade them and give students feedback. Sure, you could review solutions by hand - but that puts a massive load on the teaching team and robs students of the chance to learn from their own mistakes. So why not borrow from industry software development and go all-in on automated testing? Students get a starter template and a test suite. And then… well, then they’re adults who need to learn to read error messages and meet the spec by any means necessary.
𝗧𝗵𝗲 𝗿𝗲𝘀𝘂𝗹𝘁: a set of classic machine learning exercises with automated test-based grading. Which means anyone can try these assignments and feel just a tiiiiiny bit like a Skoltech student.
The course has already finished, and I am free to 𝗽𝘂𝗯𝗹𝗶𝘀𝗵 𝘁𝗵𝗲 𝗰𝗼𝗻𝘁𝗲𝗻𝘁 - https://github.com/fxlrnrpt/sktech_ml_homeworks_2026
𝗧𝗵𝗲𝗿𝗲 𝘆𝗼𝘂 𝘄𝗶𝗹𝗹 𝗳𝗶𝗻𝗱:
Notebooks with tasks
Helper scripts to keep the main jupyter notebooks clean
Auto-tests to provide students with immediate feedback and to automate grading
Grading script sto allow students see what grade they are going to get, prevents them to accidentally use extra files and get 0!
Pre-generated data for tests
𝗧𝗵𝗲 𝗰𝗼𝗱𝗲 𝗶𝘀 𝗽𝘂𝗯𝗹𝗶𝘀𝗵𝗲𝗱 𝘂𝗻𝗱𝗲𝗿 𝗮 𝗽𝗲𝗿𝗺𝗶𝘀𝘀𝗶𝘃𝗲 𝗹𝗶𝗰𝗲𝗻𝘀𝗲 - feel free to build upon it or re-use it in any way you want.