Where to go next

Twenty four hours is an introduction. By the end of it you should know which direction you want to go deeper in. Here is where each one leads, what it costs, and what is free.

Before you go anywhere else#

Finish the lab. Sixteen exercises, about four hours in total, and every one of them is a thing you will meet again. They are graded by tests rather than by anybody, stored in this browser and reported to nobody, so the only cost of getting one wrong is the next ten minutes.

Then build the model box. A 1:25 white card model wired as a complete show system: pixels on booms, an ESP32 taking Art-Net from a free desk on your laptop, a servo against a limit switch, a Pi playing sound on an OSC message. Between HK$900 and HK$2 600, five stages that each stop cleanly, and every layer of a real building at a scale where breaking it costs an afternoon rather than a preview.

Nothing below is worth starting before those two, because both of them will change what you want from the list.

First, the honest advice#

Build something you actually need. Not a tutorial project. The thing that annoyed you on the last production. Finishing one real tool teaches more than four courses, because only a real tool has real data, real edge cases and somebody waiting for it.

Then build the second one. The first tool teaches you the language. The second teaches you what you got wrong the first time, which is the part that transfers.

Everything below is for after that.

If you want to go deeper into programming#

The one to do first: The Missing Semester, because it fills exactly the gaps this course had to skim, in about ten hours.

If your direction is show hardware#

The step up from Class 5: learn one real-time technique properly. A hardware timer interrupt, or a state machine library, or FreeRTOS on an ESP32. Any one of them changes what you can build.

If your direction is show control and networking#

The step up: write something that speaks a protocol properly. An sACN receiver, an OSC bridge, a Companion module. Reading a specification and implementing it exactly is a distinct skill, and it is the one that makes you useful to a manufacturer.

If your direction is animation, media and content#

Check the licence of anything you build a workflow on, before the workflow exists rather than after. Remotion is free for individuals and small teams and needs a company licence above a threshold; that is a normal arrangement and it is a real line item. Finding it out at the end of a project is worse than finding it out now, and the same question applies to every tool in this table.

The step up: build a previz that reads the actual cue sheet, so the previz and the cue sheet cannot silently disagree. It is the highest-value thing in this direction for a technical director, it is a good capstone, and the principle is Class 3's rather than any one tool's: one source of truth, everything else a build artefact of it.

If your direction is AI-assisted work#

This is moving faster than any list can track, so learn what does not move.

What does not move: the verification loop, the four questions, the boundary around anything that moves or carries mains, and the fact that your name is on the risk assessment. Everything in Class 4 that is about the model will date. Everything in it that is about responsibility will not.

The step up: write your own tooling around a model. A script that reads your fixture list and answers questions about it. That teaches you more about the limits than any amount of reading.

What this course deliberately left out#

Each of these is real and worth learning, and none of them fits in twenty four hours with this audience.

Software engineering as a discipline. Object orientation, design patterns, testing strategy, concurrency, architecture. Start with A Philosophy of Software Design by John Ousterhout, which is short and better than the famous long ones.

Databases. Class 3 names SQLite and stops. Use The Index, Luke is free and is the fastest route to understanding why a query is slow.

Web development. If you need a control panel with a web interface, learn Flask or FastAPI for the back end and stop there. You do not need a framework.

Security. If anything you build touches a public network, this stops being optional. OWASP Top Ten is free and is where to start.

Machine learning itself. Class 4 teaches enough to use the tools safely. Training one is a different subject; fast.ai is the free course that respects your time.

Certifications, honestly#

There is no programming certification that matters in this industry. Nobody in live entertainment will ask whether you are a certified Python developer.

What people do ask for:

• A repository they can look at. Three finished small tools with READMEs beats any certificate.
• Something that ran in a show. With a sentence about what it did when it failed.
• The ability to read their code and ask a good question about it in the interview.

The certifications worth having in this trade are the networking and safety ones from the sibling courses, not the programming ones. Computer Systems and Networking lists those properly.

The one thing to keep#

The four questions. State, each pass, assumptions, failure.

They work on Python, on C++, on a console's macro language you have never seen, on a contractor's firmware, and on whatever a model hands you in four seconds. They will still work when every specific tool named on this page has been replaced.