We didn’t write a syllabus. We drew a map.
A syllabus tells you what comes next. It can’t tell you why. Ours is a map of 375 ideas and the 529 places they hold each other up — so before a veteran writes a line of code, they can see the whole route, where they already stand on it, and what every step is load-bearing for.
Node inspector
Pick a concept
Selecting a dot lights everything the idea rests on and prints the written reason for each link. Depth runs top to bottom: a concept sits below whatever holds it up. Nothing here gates you — it maps what a skill is made of.
Concept index
Systems & Tooling57
Engineering Practice30
Web Interface49
Programming Languages61
Data & Storage25
Services & APIs29
AI Systems58
Reliability & Operations66
What this looks like from the inside
Someone shows up having run Linux boxes for ten years and never touched JavaScript. Someone else has written Python scripts for a shop that didn’t call it engineering. Both of them can find themselves on this map on day one — not at week one of a course, but at the actual edge of what they already hold.
Then they walk the whole thing. Every veteran in a cohort covers all 375 ideas in 17 weeks, because the graph tells us exactly how much has to sit under each one, and nothing gets taught before its foundation exists.
Here’s the map by the numbers.
375
Micro-topics
One teachable idea each. Not a chapter — an idea.
529
Prerequisite links
435 that carry weight, 94 that smooth the way.
53
Domains
Grouped into 8 subjects.
16
Deepest path
35 places to start. 170 places to finish.
0
Cycles
Nothing is its own prerequisite. We check.
Four decisions, in order
Most curricula only ever answer the last one. Here's all four, and who makes each call.
The job market picks the destination, not us
We look at what employers are actually paying people to do — Lightcast demand data, O*NET, BLS wages — and let that decide what belongs on the map. When the market moves, we rebuild behind it. Nobody here teaches a thing because we happen to know it.
What a skill rests on picks the route
We broke every concept down to a single teachable idea and wired it to what it stands on. The order isn't a judgment call and it isn't a table of contents — it falls out of the map. If we can't say what an idea rests on, it isn't ready to teach.
Crawl, walk, run picks the depth
Every idea gets met three times, with less help each time: we show it, then you do it with a spotter, then you do it alone. Educators call this Gradual Release of Responsibility. Every veteran we've ever trained already knows it by the other name.
An artifact closes it out, or nothing does
Nothing gets checked off because a video ended. Each idea names the thing you have to produce and the standard it gets judged against. You built it and it meets the criterion, or you're not done. Trained to standard, or not trained.
Two kinds of link, and both say why
A line on this map isn't decoration. It's a claim about what holds what up, and every one of them carries the sentence explaining it.
Load-bearing — solid, arrowed
It carries weight
Retrieval has nothing to search until the vectors exist. Teach it the other way around and the second idea is standing on air.
Supporting — dashed, open
It smooths the way
You can build the pipeline without it. You just won’t know whether it works.
Every one of the 529 links has a sentence like that behind it. If we couldn’t write the reason, the link didn’t go in.
What one idea carries
This is the level every one of the 375 is specified to — written before the lesson, not after.
Node T20 · sample record
Measure generation faithfulness
- Concept
- Measure generation faithfulness
- Description
- Detects answers unsupported by context
- Subject
- AI Systems
- Domain
- retrieval
- Type
- meta (of: conceptual / procedural / representational / language / meta)
- Exit depth
- scaffolded (of: guided / scaffolded / unassisted)
- Market anchor
- LLM evaluation
- Evidence criterion
- Catches a fabricated answer with a scored check
- Prerequisite depth
- 14 — longest chain of load-bearing links beneath it
The line that matters is the evidence criterion. It’s the difference between a veteran who watched a lesson on measuring faithfulness and one who built a check that caught a fabricated answer. The first is a completion. The second is something a hiring manager can look at.
Type tells an instructor how to teach it. Exit depth tells them how much help to take away. Market anchor is the reason it’s on the map at all — the receipt that says an employer is paying for this.
Tap any dot above to read the same record.
Grouped by subject, sequenced by dependency
Subjects say what an idea is about. They don't say when you meet it — the links decide that, and a single subject can run the entire length of the map. Testing shows up early and never stops. So does version control.
Systems & Tooling
57 topics · 10 domains
Engineering Practice
30 topics · 6 domains
Web Interface
49 topics · 7 domains
Programming Languages
61 topics · 4 domains
Data & Storage
25 topics · 4 domains
Services & APIs
29 topics · 5 domains
AI Systems
58 topics · 7 domains
Reliability & Operations
66 topics · 10 domains
Three things a course list can't tell you
Question 01
Where does this specific person start?
A veteran who's run Linux for a decade shouldn't sit through week one of the command line. The map already knows what that decade covers and what it reaches, so we start them at the edge of it instead of the beginning of everything.
Question 02
What does this specific job actually require?
Take a real posting, name the ideas it asks for, and the map returns the set that reaches them and the order they have to come in. Not “learn React” — the eleven things under React that make React make sense.
Question 03
Why is any of this being taught?
Every idea carries the market anchor that put it there and the artifact that closes it. Every link carries its reason. When a donor asks what their giving made possible, the answer isn't hours logged — it's 375 standards met, each one named.
What we built and what we borrowed
Graph structure
Marble open taxonomy
1,590 topics, 3,221 edges, ODbL 1.0. A different subject entirely, but the architecture was right and we didn't need to reinvent it.
Market layer
Lightcast · O*NET · BLS
Skill demand, occupational profiles, and wage data. This is what sets the target.
Depth model
Gradual Release of Responsibility
Fifty years of instructional research behind the thing veterans already call crawl, walk, run.
None of these three is ours alone. Assembling them, and aiming the result squarely at what employers pay veterans for, is.
Retool. Retrain. Relaunch.
375 ideas. 17 weeks. No tuition, ever.
Vets Who Code is a veteran-run 501(c)(3). The accelerator is free, remote, and we don’t take a share of your first paycheck.
Free · Remote · 17 weeks · EIN 86-2122804