I was tired of buzzword-heavy AI projects and marginally impactful demos. Surely we can do something more inspiring with these LLMs than build another chatbot?
For me, the answer is physical AI: the moment all those breakthroughs finally reach into the real world, in robots that see, move, and figure things out for themselves. I think it is the most exciting frontier in tech right now. It is also genuinely hard to break into, because it is not one field. It is about five of them stacked on top of each other: electronics, mechanics, programming, data, and AI. Eight months ago I started my own robotics journey from scratch, and I was completely overwhelmed.
How do you get from blinking an LED to a humanoid that does your dishes?
There are thousands of scattered tutorials out there, with no sense of what comes first, or what any of it is building toward.
Stealing the best idea from my favorite games: If you have ever played a factory-building or strategy game like Satisfactory or Civ Six, you know the feeling. You start with almost nothing, and you unlock new tech one satisfying step at a time.
Those games are proof that we will happily spend hours mastering an intimidatingly complex system, as long as it is laid out as a clear tree of unlocks.
So why not point that same instinct at learning something real?
That is exactly what a tech tree is: a structured, visual path where each node is a skill and each connection is a prerequisite. You start at Curiosity on the far left and work your way right, through electronics, mechanics, code, data, and AI, all the way toward autonomous robots and humanoids. The idea is simple: turn gaming time into learning time.
Every node on the tree is a skill to learn, and the star-shaped nodes are hands-on projects where theory finally meets a soldering iron. Nodes are color-coded by discipline, so you can see at a glance whether you are in electronics, mechanics, programming, data science, or AI.
The projects are the milestones. You blink your first LED, build a little sensor station, assemble a robot arm, then a robot dog, and climb into more serious territory with things like an SO-101 arm or a Reachy Mini. The tree narrows as it feeds into each project, then fans back out again as new skills unlock on the other side.
The graph database structure allows for dependencies and visual learning paths. The content itself is built on standard documentation packages - every engineer should be more than familiar with the look!
Here is the most important part, though. I am not trying to rewrite robotics from scratch. There is already an incredible amount of brilliant content out there, from creators, courses, and docs. We are not reinventing all this knowledge. We are simply taking these amazing resources and making them easier to navigate. The tech tree is the map that sits on top of all of it and points you to the right resource for each skill, in an order that finally makes sense.
It is built to answer the three questions I hear the most:
I want to be honest: the tree is nowhere near finished. And I launched it anyway, on purpose.
It is an open source, community-driven project, because one person mapping an entire field will always miss things.
If you have a favorite tutorial, a course that finally made a concept click, or a resource you wish you had found sooner, you can add it. Open a pull request or an issue, and every node you improve helps the next person find their way. The repo has a short guide on how to add a node, and I have tagged a few good first issues to make it easy to jump in.
What happened when I hit publish
I launched quietly and shared it with a few communities I love. Within the first day, a couple hundred people were exploring the tree, dozens had cloned the repo, and the first forks showed up.
Watching the analytics, people were not just glancing and leaving. They were opening node after node, following the paths, and clicking through to resources, which is exactly the behavior I hoped the game-tree format would encourage.
It is early, it is imperfect, and it is exactly the kind of thing that gets better with more hands on it.
Come build the map with me!
If you are learning robotics, teaching it, or just curious about physical AI, I would love your eyes on it, and your contributions.
Explore the tree: [https://www.backtoengineering.com/](https://www.backtoengineering.com/)
Star and contribute: [https://github.com/iuliaferoli/backtoengineering](https://github.com/iuliaferoli/backtoengineering)
What would you add to the tree? Let me know in the comments, or better yet, in a PR.