Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Duration 14 hours
Course Outline
Refresher on AutoGen Core Concepts
- Defining agents and group configurations
- Mechanisms of function calling and role chaining
- Identifying limitations in built-in agents and the necessity for customization
Constructing Custom Agents with Python
- Defining agent behavior through user_proxy and AssistantAgent subclasses
- Embedding role-specific logic and decision-making processes
- Developing reusable agent modules and mixins
Advanced Tool Integration and Routing Strategies
- Processes for tool registration, binding, and execution
- Implementing conditional routing of inputs to designated tools
- Orchestrating multi-step toolchains and composite actions
Planning and Context Management Frameworks
- Architecting task decomposers and intermediate planners
- Ensuring context continuity across chained agent interactions
- Deploying scoped memory systems for extended sessions
Robust Error Handling and Recovery Protocols
- Identifying and mitigating failed or incomplete interactions
- Implementing agent-initiated retries and fallback logic
- Comprehensive logging, debugging, and response validation
Multi-Agent Collaboration with Custom Roles
- Coordinating specialized agents within dynamic groups
- Orchestrating reasoning loops and cooperative workflow structures
- Evaluating role separation versus role blending in task distribution
Strategies for Real-World Deployment
- Optimizing performance and cost efficiency (token usage, caching)
- Integrating AutoGen workflows into web applications or data pipelines
- Addressing security, observability, and user feedback integration
Requirements
- Strong proficiency in Python programming.
- Practical experience in building LLM-based applications.
- Solid understanding of function calling and multi-agent system architecture.
Target Audience
- Senior Developers
- Platform Engineers
- AI Architects
Testimonials (1)
I liked that he constantly provided examples but also offered time for individual work on what he presented.