Thoughts about a system which suggests you project ideas and potential course of action based on your background and preferences.

Problems⌗

Not enough space⌗

From personal experience one of the more efficient ways to learn is by building something from scratch. Most of the books and courses I came across either not provide practice at all, or provide it by the means of exercises, or provide it by walking you through the learning project, not giving you enough space for your own thoughts and experiments. Even if you try to go through the provided learning project by yourself, your solution quickly diverges from the author’s, and it becomes difficult to follow the course.

Coming up with a project idea⌗

That’s often the case that you want to learn a new technology, either to apply for a job or to stay relevant as a professional or just for the fun of it, but don’t have any particular project on your mind. You may want to learn a new programming language or dive into an entire field like bioinformatics, and you don’t know what would be a good project to start with, but you may have a vague idea of what you’d like it to be about. Say, you were really fond of taxonomy in high school, or you were always fascinated by compilers. A relevant project idea can be a motivator on your learning journey.

Solution⌗

Tell what and how you’d like to build⌗

You enter a prompt, which can be as vague as how to make a website or as precise as how to train a diffusion model using PyTorch, and you get a set of cards corresponding to specific learning projects. You are free to expand the project cards and learn more about each of them, before you choose the most appealing to you. Each project guides you through 2 dimensions:

  • domain knowledge required to build the project;
  • knowledge of developer tools and approaches.

For example, it can suggest you to build a multi-coin crypto wallet in Flutter, a mobile game of chess in Swift, a reinforcement learning neural net playing Mario in Python.

Choose appropriate level of detail⌗

Each project is a step-by-step guide where you can vary the level of detail of each step to provide yourself with a challenging yet approachable task:

  • if you feel like the guide tells you too much and gives you too little space to think about the problem on your own, you decrease level of detail and get more involved with the task;
  • if you’re stuck on a problem for hours or days, you increase level of detail and get a hint.

One way to think about it could be a gym: if you’re too easy on yourself you won’t get expected results, and if you try too hard you can either get an injury or inflict on yourself as much pain as to despise the whole enterprise.

[bonus] Provide bio⌗

Free-form description of your past experiences can be a useful starting point for the app to both suggest you new projects and to choose the initial level of detail.

Users⌗

STEM⌗

Programming is the main target, but I hope to add mathematics and theoretical physics too, replacing project steps with exercises. Experimental physics could be added given the project is build inside of a software simulator.

Somewhat experienced students and mentors⌗

I believe that a lot of people who are just starting out in the field need more than a guide: they need a mentor, and I feel like a mentor boils down to AGI. So in this case I’d rather concentrate on being a useful tool for humans mentors. In case of highly motivated beginners and somewhat experienced users though, this app should be useful by itself.