Capstone: Build, Document, and Present an AI-Powered Project

The Purpose of the Capstone

The capstone is not a test. It is a translation. It takes the understanding and skills you have built across this path and translates them into a single artifact that demonstrates AI fluency at the level of an informed practitioner.

The artifact is not impressive because of its technical sophistication. It is impressive because it shows you can frame a problem, build something that addresses it, evaluate that something honestly, and communicate the results clearly.

Choosing Your Capstone Project

Review the three capstone options in the path description. Select based on what role you are targeting:

Capstone A (AI-Powered Tool): Best if you are moving toward building AI products or applied AI engineering. Shows you can go from idea to deployed tool.

Capstone B (Data Story): Best if you are moving toward data analysis, AI product management, or operations. Shows you can turn data into decisions.

Capstone C (AI Tool Audit): Best if you are moving toward AI evaluation, trust and safety, or governance. Shows you can evaluate AI systems systematically.

For any of these: choose a domain you know well. A capstone about a problem in your professional context is more impressive and more useful than a generic tutorial project.

The Four Deliverables

Every capstone produces four artifacts:

1. Working demo or documented output: Something that runs or something that can be shown. For Capstone A: a deployed application or a recorded walkthrough. For Capstone B: a Jupyter notebook with visualizations. For Capstone C: a structured evaluation report.

2. Written evaluation: What you tested, what you measured, what success rate you achieved, what failure modes you found, and what you would improve next. This must be honest - including failures makes it more credible, not less.

3. Non-technical summary: Two pages maximum. Explains what you built, why, and what it shows - to someone who has not taken this path and does not know the vocabulary. This tests whether you truly understand what you built.

4. Five-minute presentation: You should be able to present your capstone to any audience in five minutes. Practice this out loud. The ability to explain a technical project clearly and concisely is one of the most valuable professional skills you can demonstrate.

Common Capstone Mistakes

Scope too large: "I am going to build a full RAG system over my company's entire document library." Start smaller. A working tool over 20 documents with documented evaluation is a stronger capstone than an ambitious project that was never finished.

No evaluation: The most common mistake. "I built it and it works" is not a capstone - it is a demo. Show the numbers.

Hiding failures: A capstone that says "it worked great" is suspicious. A capstone that says "it achieved 82% accuracy on my test set, failed most often on [specific input type], and I would fix it by [approach]" is credible and shows engineering judgment.

Choosing an unfamiliar domain: The capstone demonstrates AI fluency applied to a domain. If you do not know the domain, you cannot evaluate whether the AI output is actually good.

The Capstone as a Career Artifact

A well-documented capstone is the most effective career artifact you can produce from this path. It demonstrates:

• You can apply AI tools to real problems, not just use them casually
• You evaluate what you build rather than just deploying and hoping
• You can communicate technical work to non-technical audiences
• You have completed something, end to end, independently

Link to it in your LinkedIn profile. Reference it in job applications and interviews. It is the concrete evidence behind the abstract claim "I have AI fluency."

You Have Completed This Path

By finishing this capstone, you have:

• An accurate mental model of how AI systems work
• Practitioner-level use of AI tools with systematic evaluation discipline
• Basic Python and data literacy that enables independent data work
• Experience building and evaluating an AI application
• The vocabulary and frameworks to contribute to AI product and operations discussions
• A concrete artifact demonstrating all of the above

AI fluency does not mean you can train a neural network. It means you can think clearly about AI, use it well, build with it, evaluate it honestly, and communicate about it to any audience. You can do all of that now.

The next step is the one you defined in Module 23: the role you are targeting, and the six-month plan you wrote to move toward it. Start now.