Build to Certify
Hands-on preparation for the Claude Certified Architect exam.
Build to Certify is not a survey of AI. It's not a prompt-writing workshop. It's eight weeks of structured work that maps directly to the five domains of the Claude Certified Architect exam — and exits with a portfolio that qualifies you for exam registration through the Anthropic Partner Network.
Each module pairs a video walkthrough with runnable Jupyter notebooks and start-up Python scripts so learners focus on judgement rather than scaffolding.
About the Claude Certified Architect (CCA)
The CCA is Anthropic's professional certification for engineers who build production AI systems with Claude. It's a proctored, 120-minute, 60-question exam covering five domains — and unlike most AI credentials, it tests whether you can architect real systems, not just write prompts. Build to Certify maps directly to all five domains.
| Domain | Topic | Weight |
|---|---|---|
| 1 | Agentic Architecture | 27% |
| 2 | Tool Design / MCP | 18% |
| 3 | Claude Code | 20% |
| 4 | Prompt Engineering | 20% |
| 5 | Context Management | 15% |
The eight-week curriculum
Build to Certify is structured as eighty modules, each mapped to a CCA exam domain and culminating in a MAKE artifact you ship to your portfolio. The cohort takes one module per wee. Self-paced students move at their own speed but cover the same content and exit with the same portfolio.
| Week | Module | Domain | Weight | Focus & Exam Connection |
|---|---|---|---|---|
| 1 | Claude API Foundations | D4: Prompt Eng. (partial) | 20% | Messages API, system prompts, stop_reason, tokens, temperature. The request/response foundation every later week builds on. Includes GitHub onramp. |
| 2 | Prompt Engineering & Structured Output | D4: Prompt Eng. | 20% | Few-shot examples, tool_use schemas, Pydantic validation, retry loops. Targets the exam's most-missed D4 question: programmatic enforcement vs. prompt requests. |
| 3 | Context Management & Reliability | D5: Context Mgmt. | 15% | Context windows, RAG strategies, lost-in-the-middle, prompt caching, progressive summarization. Measured tradeoffs across the 'small but tricky' domain. |
| 4 | Tool Design & MCP — Part I | D2: Tools & MCP | 18% | Tool schemas, descriptions as selection signal, tool_choice modes, structured errors. Demonstrates selection failure before the fix — the exam's preferred shape. |
| 5 | Tool Design & MCP — Part II | D2: Tools & MCP | 18% | MCP architecture, tools vs. resources vs. prompts, mcp.json, server scoping, error contracts. Students stand up a working MCP server end-to-end. |
| 6 | Agentic Architecture — Part I | D1: Agentic Arch. | 27% | The agentic loop from scratch: stop_reason branching, history injection, programmatic prerequisites, hooks. Every loop failure mode the exam tests. |
| 7 | Agentic Architecture — Part II | D1: Agentic Arch. | 27% | Hub-and-spoke orchestration, subagent isolation, structured recovery, deterministic escalation. The course's deepest week — and the exam's heaviest domain. |
| 8 | Claude Code + CI/CD + Exam Sim | D3: Claude Code | 20% | CLAUDE.md hierarchy, slash commands, plan mode, hooks, the -p flag, Agent Skills. Closes with a timed practice exam and anti-pattern review. |
Who this course is for
This course is for you if:
You can write basic Python (functions, loops, dictionaries) but haven't built with AI APIs yet
You want a credential that maps to where AI engineering jobs are actually moving
You learn best by making things, not by watching survey videos
You're ready for eight weeks of real work, not a weekend bootcamp
Not yet, if:
You've never written code in any language
You're looking for a tour of AI tools rather than building real systems
You want to learn prompt engineering specifically — there's a domain for that here, but it's one of five
Two ways to take the course
Same content. Same eight modules. Same MAKE artifacts. Same exit portfolio. The difference is pacing and community.
Cohort — June 15, 2026
Eight-week guided experience with a defined start and finish
Weekly cadence keeps you accountable and on schedule
Cohort group for questions, peer feedback, and shared portfolio review
Live office hours with drC during the cohort program
Best if you work better with structure and a real deadline
Self-paced — Start anytime
Same eight modules, same MAKE artifacts, same portfolio outcome
Move at the speed your life allows — finish in six weeks or sixteen
Lifetime access to course materials and updates
Email access to drC for portfolio review and key questions
Best if you need flexibility around your existing commitments
Ready to start?
The next cohort begins June 15, 2026. Self-paced enrollment is open year-round. Either way, you exit with a portfolio that proves you can build production AI systems — and a clear pathway to CCA certification.
A note on CCA exam access
The Claude Certified Architect exam is administered by Anthropic through their certification program. The AI Edge has applied to the Anthropic Partner Network — the affiliation that supports learners through to exam registration. Course completers will be supported in accessing the CCA exam through that affiliation as it becomes formally established.
What this means in practice: the curriculum, the artifacts, and the portfolio you build are designed to prepare you for the exam regardless. When the affiliation is in place, course completers will receive direct registration support. We're committed to building Build to Certify alongside the Anthropic ecosystem, not separate from it.