Week 2 · Thu, Sep 24, 2026 · Open models, model selection, and API basics
Pick the right AI model for the job
Stop collecting models and start building a decision process. We scored real tasks on three questions, treated every model as a plain web endpoint, and fixed a router that missed a deadline buried in a community board notice.
What we covered
Score the task before you pick a model
The Three C's: Complexity (how hard is it?), Context (what does it need to know?), and Criticism (how bad is a wrong answer?).
Every model is just a web endpoint
A request goes in, a response comes out. Open-weight doesn't mean private: a hosted open model still runs on someone else's server.
Match the model to the risk
Small model for low-stakes work, bigger model when mistakes are costly. Treat that as a guess you test, not a law.
Spec first, then prompt
A heavy model plans, a light model does the work, a second model checks it, and you make the call.
Code that runs isn't code that's done
Green checkmarks can sit on top of wrong data. Check the output against the source.
Try the homework
15–20 minutes required, optional levels after. A free GitHub account and a browser. Nothing to install. Only Level 3 needs an AI API key.
- On the repo, click Use this template to make your own public copy. The Actions tab starts with a red X on purpose.
- Level 1 (required): fix extract_event_date() in extractor.py so the bot books the show, not the presale lottery. Then answer decision.md.
- Level 2 (challenge): rewrite route() in router.py to send each message to rules, a model, or a human.
- Level 3 (bonus): send model-tier messages to a light model first and escalate to a heavy one only when needed, in escalate.py.
Fellows turn in homework through Google Classroom. The Classroom link is on the last slide.
Want the full fellowship?
Classes are open to everyone. Fellows also get homework feedback in Google Classroom, a cohort Discord role, office hours, and priority seats at the hackathon.