Other certification · CCAR-P

Claude Certified Architect — Professional (CCAR-P)

Overview guide for the Claude Certified Architect — Professional exam — the advanced tier above the Architect Foundations exam this repo covers in depth. Facts below are taken from the official Exam Guide v1.0 (effective July 2026).

Preparing for Foundations first? Start with the full CCAR-F study guide — Anthropic's Professional blueprint assumes you operate comfortably at that level.

What It Validates #

That you can design, build, and deliver production-grade AI solutions on the Claude platform as an architect who owns the full lifecycle: selecting models, architectures, and API patterns; prompt and context engineering; integrating Claude into enterprise systems; and incorporating evaluation, security, compliance, and governance into designs — plus the distinctly Professional-level skill of stakeholder communication and lifecycle management.

Intended for: mid- to senior-level solution architects, AI/ML engineers, and technical leads. Recommended (not required): 3+ years in systems architecture or platform engineering, 6+ months hands-on with Claude (or comparable LLM systems) in production, experience delivering end-to-end systems from discovery through operationalization.

Not intended for: entry-level developers, casual users, or roles limited to isolated tasks (e.g. prompt writing) without broader system-design responsibility.

Exam at a Glance #

Domain Blueprint #

Seven domains vs Foundations' five. The Foundations themes reappear at higher altitude, and two areas are effectively new: RAG-heavy Integration, and Stakeholder Communication & Lifecycle Management.

Key objectives per domain #

1. Solution Design & Architecture (17%) — translating business problems into Claude-based solutions; end-to-end architectures (input → processing → output → feedback loops); choosing among workflow, agentic, and augmented-LLM patterns; multi-agent systems and orchestration; decomposition; aligning to business value pillars (efficiency, cost, performance SLAs).
2. Claude Models, Prompting & Context Engineering (13%) — model selection by trade-off; system prompts, templates, guardrails; zero-shot/few-shot/chain-of-thought; context-window and token optimization; prompt reuse (caching, modular prompts, Skills).
3. Integration (19%) — the heaviest domain: tool/agent capability-bloat evaluation, authn/authz gap analysis, accuracy-latency trade-offs, observability at scale, RAG pipeline design (chunking, indexing, retrieval matched to data shape and query pattern), choosing integration mechanisms (MCP vs API/CLI vs agent-to-agent), progressive-discovery vs monolithic context strategy.
4. Evaluation, Testing & Optimization (16%) — evaluation metrics (accuracy, latency, cost, safety, security), eval datasets and test frameworks, A/B testing, diagnosing prompt failure/hallucination/model mismatch, token/latency/cost optimization, logging and observability.
5. Governance, Safety & Risk Management (14%) — guardrails and safety controls, LLM failure modes, human-in-the-loop validation, regulatory compliance (GDPR, HIPAA, FedRAMP), ethical AI (bias, fairness, transparency).
6. Stakeholder Communication & Lifecycle Management (14%) — structured discovery and requirements gathering, communicating architectural decisions and trade-offs, feedback loops and expectation/SLA alignment, architecture documentation, supporting lifecycle phases (discovery, design, handoff, monitoring, iteration). This domain does not exist on CCAR-F.
7. Developer Productivity & Operational Enablement (7%) — configuring Claude tooling for teams (e.g. Claude Code), improving developer workflows with AI-assisted tooling, supporting debugging and operational issues.

Sample-question flavor (from the official guide) #

The published samples test least-privilege tool configuration (remove unneeded refund/delete tools rather than log or confirm), prompt-caching architecture (stable prefix first + caching to cut latency and cost), and RAG debugging (confident-but-wrong answers after a document refresh → suspect retrieval/indexing first). Expect judgment-under-tradeoffs questions, not recall.

Relationship to CCAR-F and This Repo #

• All five CCAR-F domains reappear inside CCAR-P domains 1–5 — the existing deep-dives remain the right base layer.
• Study the deltas on top: RAG architecture (chunking/indexing/retrieval strategy), evaluation frameworks and A/B testing, regulated-industry compliance, and stakeholder communication / lifecycle management — none of which the Foundations guide covers today.
• Anthropic recommends sitting CCAR-F first unless you already have deep, current production experience.

How to Prepare (per the official guide) #

• Study the blueprint and self-assess against each objective
• Review official documentation for the Claude API, models, prompt engineering, MCP, and Skills
• Build and operate at least one end-to-end Claude solution including RAG, evaluation, and observability
• Practice architectural decision-making: model selection, integration protocols, security trade-offs

Registration and Policies #

1. Register via the exam's page on the Anthropic Partner Academy (partner-tier discounts apply at checkout), then schedule through Pearson VUE — online proctoring or a test center.
2. Cancel or reschedule free of charge up to 48 hours before the appointment (a cancellation in that window is fully refunded — email Anthropic certification support, since cancelling in Pearson VUE alone doesn't trigger the refund); changes within 48 hours forfeit the fee. The v1.0 Exam Guide still says 24 hours; the live Partner Academy policies page is newer.
3. Retakes: up to 4 attempts per rolling 12 months, with waiting periods after each failed attempt (14 / 30 / 90 days), per Pearson VUE program policy.
4. Renewal: the credential is valid 12 months; renewing on time is free — review what changed since you certified and pass a non-proctored, open-book assessment (retakes allowed; the new 12 months run from your current expiry). If it lapses, the full exam fee applies again. Registration requires a partner-domain email address. Delivery moved to Pearson VUE and digital badging to Credly on 2026-06-30; candidates must be at least 18. Partner-tier discounts apply at checkout — 50% for Select, Preferred, and Global Premier partners, 100% for Global Premier partners through 2026-12-31. The FAQ lists the exam and its prep content in English and Japanese (one registration; you pick the language when scheduling), and a Japanese translation of the v1.0 exam guide was published on 2026-10-08.

Resources #

• Official CCAR-P Exam Guide (PDF, v1.0)
• Pearson VUE — Anthropic certification program
• CCAR-F study guide in this repo — the base layer