Anthropic Claude Certified Developer Foundations (CCDV-F) Cert Prep
8h 9mIntermediate2026-09-15
Authors
Tutorials Dojo
Course details
Examine the intricacies of developing AI applications using Claude. Learn about crucial elements such as the hub-spoke model for task distribution, the agentic loop for efficient AI workflows, and multi-agent systems that improve task complexity management. Discover how to use Claude Messages API efficiently, handle real-time and batch data processing, and refine your prompt engineering. Explore essential security features including application security and secret management, to ensure reliable and secure deployments. The course covers different tool use cases, architectural strategies for agentic AI, and model selection. It is designed for AI and machine learning engineers looking to deepen their understanding of development protocols. By the end of the course, you will be equipped to create efficient, secure, and scalable AI applications using Claude features.
Concepts
Introduction
- Claude Certified Developer Foundations (CCDV-F) overview
Agents and Workflows
- Introduction to the hub-spoke model
- What is the agentic loop
- Introduction to tool call interception hooks
- How to branch on tool use
- How a Coordinator Decomposes Tasks
- Choosing an agent framework (Strands vs. LangGraph vs. Pydantic AI)
- Agentic AI vs. generative AI
- The architecture of agentic AI
- Types of agentic AI
- Multi-agent systems - When AI agents work together
- Memory and context in AI agents
- What is Amazon Web Services (AWS) Strands Agents
- What is LangGraph
Requirements and Lifecycle
- What are functional vs. infrastructure requirements
- What is the software development lifecycle (SDLC)
- Gathering functional and infrastructure requirements
- Managing a Claude app across the SDLC
The Claude Messages Application Programming Interface (API)
- What is the Claude Messages application programming interface (API)
- What are Messages API data access patterns
- What is streaming in the Claude API
- What are vision and multi-format inputs
- What is extended thinking
- Working with the Messages API - Tools, streaming, vision, and thinking
Delivery, Cost, and Access Paths
- What is prompt caching (API)
- What does invoking Claude through third-party vendors mean
- When to use the batch API
- When to use the synchronous API
- Choosing real-time vs. batch API
- Integrating Claude via software development kits (SDKs) - error handling and multi-format input
- What is the Message Batches API
Software Engineering Foundations
- What is version control
- What is code review
- What is refactoring (small- and large-scale)
- What is a representational state transfer (REST) API
- What is JavaScript Object Notation (JSON)
- What is asynchronous programming
Claude Application Design
- How does Claude interpret instructions across interfaces
- What are content boundaries
- What is schema design
- What is session hygiene
- What are plugin management and plugin dependencies
- Designing a Claude application across interfaces
- Designing schemas and content boundaries
Configuration Management
- What is a CLAUDE.md file
- What is settings.json
- What is model version pinning
- What is prompt versioning
- Managing configuration - CLAUDE.md, settings.json, version pinning, and prompt versioning
Claude Code
- What is Claude Code
- What are Claude Code's core components
- What are slash commands
- What are headless, streaming, and auto modes
- How to run Claude Code in continuous integration and continuous delivery (CI CD)
- What the -p flag does
- Introduction to the CLAUDE.md Hierarchy
Eval, Testing, and Debugging
- What is an eval
- What are the common error types in Claude apps
- What is trace analysis
- What is a failure mode
- Designing and running an eval
- Monitoring production quality
- Debugging - isolating integration-layer vs. model-output problems
- Selecting an error-recovery strategy
- What is LLM as a Judge
- What is Langsmith
- Understanding Model Drift in Machine Learning
Large Language Models (LLMs) and Technical Fundamentals
- What is a token
- What is a context window
- What is sampling (temperature top-p)
- What is nondeterminism in LLMs
- What is next-token generation
- Model processing mode options - Fast mode, extended thinking, adaptive thinking, and effort levels
- What are zero-shot, single-shot, and multi-shot prompting
- What is a WebSocket
- What are LLMs
- What are Foundational Models
- Neural networks - Basics and types
Model Selection and Trade-Offs
- What are Opus, Sonnet, and Haiku (and their use cases)
- What are breaking behavior changes across model releases
- Choosing a model - Haiku vs. Sonnet vs. Opus
Cost and Token Management
- What is token usage tracking and cost modeling
- Prompt Caching in Claude
- How to Choose Between Cost and Latency
Context Engineering
- What is tool output pruning
- What is compaction
- What is context isolation
- Managing context to prevent drift and bloat
Prompt Engineering
- What is prompt engineering
- What is system vs. user prompt placement
- What is an output constraint
- What is input sanitization
- Writing and iterating effective prompts
- Placing instructions across components (system vs. user)
Output Handling
- What is structured output
- What are response validation and defensive parsing
- Producing and validating structured output
Application Security
- What is prompt injection
- What are data sensitivity and Personally Identifiable Information (PII)
- Defending against prompt injection and jailbreaks
- What are authentication, authorization, confidentiality, privacy, and integrity
Guardrails, Hooks, and Secrets
- Introduction to PostToolUse hooks
- What are secrets and API key management
- Layering guardrails and applying least privilege
- Using hooks to block destructive actions
- What are content policy and guardrail layering
- What is secure by design (SbD)
- What is a Claude hook
Tools and Model Context Protocols (MCPs)
- Tool use in Claude
- How to Write a Complete Tool Description
- What is a programmatic gate
- Picking the right tool - Skills compared to everything else
- Understanding how skills are created and work
- What is an approval pattern
- What is a Claude skill
- What is a Model Context Protocol (MCP) server
- Agent Communication Protocol (ACP)
- Agent-to-agent (A2A)
- What are MCP resources, tools, and prompts