Get in Touch
 Duration 14 hours

Course Outline

Introduction

  • Differences between workflows and agents
  • Overview of architecture and common application scenarios

Basics of the Claude API

  • Messages API, available models, and configuration parameters
  • System prompts and structured output generation
  • Streaming mechanisms and error management
  • Lab: Initial API integration exercise

Tool Integration

  • Creating tool definitions and schemas
  • The tool-execution cycle and strategies for error recovery
  • Lab: Developing an agent that interacts with internal company APIs

MCP and Agent Skills

  • Overview of MCP servers and clients
  • Leveraging pre-existing servers and developing a custom one
  • Utilizing Agent Skills for reusable capabilities
  • Lab: Linking an agent to a specific data source

Claude Agent SDK

  • Agent loops, session management, and context control
  • Managing permissions, hooks, and subagents
  • Lab: Creating an agent for research and reporting tasks

Cost, Quality, and Safety

  • Prompt caching techniques and token budget allocation
  • Implementing guardrails and human approval mechanisms
  • Conducting evaluations using a limited test dataset
  • Logging processes and monitoring strategies

Transitioning to Production

  • Deployment strategies and secure secret management
  • System maintenance and version control
  • Creating a roadmap for your specific use case

Requirements

  • Proficiency in programming with Python or TypeScript
  • Practical experience in interacting with REST APIs and processing JSON data
  • Foundational knowledge of large language models

Target Audience

  • Software developers and data engineers integrating AI functionalities into products or internal tools
  • Solution architects assessing agent-based architectures
  • Technical teams seeking to transition from prototype stages to production environments

Number of participants


Price per participant

Testimonials (3)

Upcoming Courses

Related Categories