Courses

Prepare for the Claude Certified Architect – Professional (CCAR-P) exam (63 questions, 120 minutes, pass at 720 of 1,000). Forty hours cover all seven exam domains, weighted the way the exam weights them.

• 20 two-hour modules: solution design (17% of the exam), models and prompting (13%), integration (19%), evaluation (16%), governance (14%), stakeholder communication (14%) and developer productivity (7%), then a capstone, the mock final and a score review.
• 37 labs on one case study. You design, build, evaluate and defend a Claude support and document-intelligence platform for Meridian Mutual, a financial-services firm. Incident labs hand you a broken system to fix: an over-privileged agent, a cost-bleeding uncached prompt, a RAG index that turns confidently wrong after a refresh.
• 342 exam-style questions, each with a rationale for the right answer and every distractor: a 21-item diagnostic, 7 domain pre-checks, 17 module quizzes, 7 domain exams, a timed 63-item mock final and four 15-item sets for the 30 days before your exam.
• Role-plays for the soft domains: a discovery interview with a COO, an architecture decision defended to a skeptical CISO, an SLA negotiated with a CFO.
• A portfolio you keep: Architecture Decision Records, architecture diagrams, an evaluation report and a capstone presented to a mock architecture review board.

Prepare for the Claude Architect Foundations (CCA-F) exam with 37 self-paced lessons in nine modules, covering all five exam domains.

• Agentic architecture and orchestration: the agentic loop, ending it on stop_reason with iteration caps, hub-and-spoke coordinators, task decomposition, subagents, and session resume and fork in the Agent SDK.
• Tool design and MCP: tool descriptions and JSON Schema inputs, MCP tools vs. resources vs. prompts, server discovery and MCP permission rules.
• Claude Code configuration: settings scopes, deny/ask/allow permission rules, permission modes, sandboxing, session commands and GitHub Actions.
• Prompting and structured output: few-shot examples, tool_choice, schema-enforced output, validation-retry loops and fresh-context review passes.
• Context management and reliability: large-context failure modes, scratchpads and external memory, provenance, the Message Batches API and escalation to a human.
• Practice: a 15-item diagnostic, five domain knowledge checks and a timed 38-question practice exam at two minutes per question, with a rationale for every option.