# SkillTree Sims — AI Roadmap Progression Simulator (SimPy)

## PURPOSE
The video insists that learning AI is a map, not a line: everyone clears Tutorial Island,
then commits to one class skill tree before touching the next. Each class has a fixed
skill tree — the Strategist path runs vendor landscape to running AI teams to setting
strategy to leading adoption; the Rogue goes from "what agents can do" to repeatable
workflows to a personal agent setup; the Alchemist moves from agent fundamentals through
evals and reliability to production; the Sage climbs from how models work through training
neural networks to fine-tuning and inference; the Machinist goes from hardware basics
through quantization to scaling and robotics.
This project simulates that progression. Given a class and a weekly study budget, it
reports how many nodes the learner clears, in what order, and how many weeks it takes.

## WHY THIS MATTERS
It makes the transcript's dependency graph executable. In the video, skills unlock other
skills — you cannot "lead AI adoption" without first understanding the vendor landscape,
and you cannot fine-tune before you understand training. Modeling those prerequisites as
a discrete-event graph shows students why skipping ahead stalls a roadmap and why the
same number of study hours yields different results per class.

## HOW TO BUILD
Stack: **SimPy + NumPy-free plain Python**. Represent each class as `{node: (hours,
[prereqs])}`, spawn one SimPy process per node, and have each process wait until its
prerequisites land in a shared `done` set before consuming its study time.

```python
def run_node(self, node_id, hours, prereqs):
    while not all(p in self.done for p in prereqs):
        yield self.env.timeout(1)              # blocked on prerequisite
    yield self.env.timeout(hours / self.weekly)
    self.done.add(node_id)
```

A seeded `random.Random` adds ±15% jitter so runs are reproducible but not flat.

## WHEN TO USE IN PRODUCTION
Use it to plan curricula: which class is realistic in 12 weeks at 8 hours/week, where
learners stall, and how many hours a certificate actually implies. It is also a template
for any prerequisite-gated onboarding. Do NOT use it for real scheduling (SimPy models
time, not calendars), for large arbitrary graphs where a topological sort suffices, or
when you need wall-clock dates rather than relative weeks.

## HOW IT HELPS STUDENTS
Students learn discrete-event simulation, dependency modeling, and reproducibility with
seeded RNGs — a transferable pattern for capacity, pipeline, and queue problems.

## STARTER CODE
`starter/main.py` defines the Tutorial Island pack plus all six class skill trees, a
`Learner` actor, and a `simulate()` entry point, with `# TODO` markers for SimPy
resources and quest rewards.
