No tutorials. No hand-holding. You decide what to build — and then you build it.
Day 76 of 80
Every previous week had a clear path: here's the concept, here's the pattern, here's the exercise. That structure was intentional — it kept the learning tight and the feedback fast.
Week 16 is different. There are no tutorials to follow. No exercise with a known solution. You choose what to build, you write the plan, and you build it from scratch over the next five days.
This is what real development looks like. The skills you've been building for 15 weeks — Claude API calls, async code, CLI tools, databases, testing, Flask apps — all of it comes together here. You already know how to do each piece. This week is about assembling them into something whole.
Day 76 is for choosing. Day 77 is for planning. Building starts on Day 78. Projects that skip planning stall out halfway through — not because the builder lacks skill, but because they run into a structural problem they didn't see coming. Give the plan the time it deserves.
Choose one. Each is scoped to five days of focused work. Each uses real skills from this course in combination.
A CLI tool that reads a client brief — a plain text file or PDF — extracts shot descriptions using Claude, generates platform-specific prompts for each shot in parallel, and exports a formatted shot list as both JSON and HTML.
The full loop: brief in, prompts out. One command. One deliverable.
python shot_processor.py brief.txtshots.json (machine-readable) + shots.html (shareable with a client)Difficulty: Moderate. Skills used: Week 4 (Claude API), Week 5 (modules), Week 13 (async), Week 15 (Typer CLI).
Add a feature to INFLECT — the AI filmmaking tool you've been building throughout the course. You now understand Flask, the Claude API, databases, and the existing codebase. Find something to add or meaningfully improve, and ship it.
This option proves something different from the other two: that you can contribute to an existing project. That's 90% of professional software work. Reading unfamiliar code, understanding its conventions, and extending it without breaking it — that's the skill.
Difficulty: Variable — depends on the feature you choose. Skills used: All weeks.
A FastAPI + HTML dashboard that becomes the control center for your AI filmmaking workflow. It stores all your generated prompts, lets you organize them by project and scene, export batches for a shoot, and track which prompts you've actually used.
Difficulty: Moderate to Hard. Skills used: Weeks 11–14 (databases, FastAPI, templates).
If you're genuinely torn, answer these three questions. Your answers usually point clearly at one option.
| # | Question | Why it matters |
|---|---|---|
| 1 | Which one would you actually use after this course? | If you'd use it, you'll finish it. If you won't, you'll lose steam on Day 78. |
| 2 | Which one excites you more than the others? | Mild technical interest isn't enough to carry you through a build. Pick the one that makes you want to start. |
| 3 | Which one plays to your strengths from this course? | Not what you find easiest — what you find most satisfying. Hard and rewarding beats easy and boring. |
All three options are scoped correctly for five days. All three use real skills. All three produce something you can actually show someone. The worst decision you can make is spending today paralyzed between options instead of committing to one and starting to plan.
Create a note called Capstone Project Plan in your Obsidian vault. It doesn't need to be long — it just needs to exist and have real answers. Write:
Day 77 walks through a full planning template — file structure, data model, the hard part, and a map of which week's skills each feature uses. Bring your chosen project. You'll leave Day 77 with a complete plan and be ready to build on Day 78.