Projects and code implementations from the DeepLearning.AI course "Design, Develop, and Deploy Multi-Agent Systems with CrewAI" (December 2025).

This course, taught by João Moura (Co-founder and CEO of CrewAI), provides comprehensive training on building production-ready multi-agent systems that automate complex, end-to-end workflows. The course bridges the gap between prototyping AI agents and deploying them at scale with reliability and control.

By completing this course, you will:

• Build multi-agent systems that plan, reason, and collaborate to solve complex problems
• Design intelligent agent teams using tools, memory, and guardrails
• Implement real-world design patterns for agent coordination and collaboration
• Monitor and debug agent performance using traces and observability tools
• Evaluate agent behavior with LLM-as-a-Judge techniques
• Apply best practices for scaling agents from prototype to production

This repository contains notebooks and projects organized by course modules:

├── Module1/  # Design and Development Fundamentals
├── Module2/  # Agent Control and Tool Integration
├── Module3/  # Advanced Orchestration and Monitoring
└── README.md

Building Single- and Multi-Agent Systems from Concept to Prototype

This module covers the foundational concepts of designing and developing agent systems:

• Creating single-agent and multi-agent architectures
• Tuning agent behavior using context engineering
• Understanding agent roles, goals, and backstories
• Examining real-world use cases and production deployments
• Practical Application: Automated Code Reviewer system

Key Topics:

• Agent system architecture and design patterns
• Context engineering for agent behavior
• Production considerations and case studies
• Role-playing and task delegation

Controlling Behavior with Guardrails, Memory, and Tools

This module focuses on enhancing agent capabilities through advanced control mechanisms:

• Implementing guardrails to ensure safe and predictable behavior
• Using execution hooks for fine-grained control
• Adding memory systems (short-term, long-term, and shared)
• Building and integrating custom tools
• Understanding the Model Context Protocol (MCP) for expanding agent capabilities

Key Topics:

• Guardrails and error handling
• Memory architectures for agents
• Custom tool development
• Model Context Protocol (MCP) integration
• Knowledge management for richer decision cycles

Complex Coordination Patterns and Production Monitoring

This module explores sophisticated orchestration strategies and production-ready monitoring:

• Orchestrating agents using sequential, parallel, hierarchical, hybrid, and asynchronous patterns
• Implementing Flows as a low-level control layer
• Monitoring multi-agent systems with tracing and observability tools
• Training agents using human-in-the-loop feedback
• Conducting structured evaluations with LLM-as-a-Judge
• Practical Application: Automated Code Review Flow with advanced orchestration

Key Topics:

• Sequential, parallel, and hierarchical coordination
• CrewAI Flows for low-level control
• Tracing, sampling, and observability
• Human feedback integration
• LLM-as-a-Judge evaluation techniques

Industry Adoption and Production Patterns

This module examines how multi-agent systems are deployed in production environments:

• Analyzing business adoption across industries and functions
• Learning from chatbots to workflow co-pilots evolution
• Studying real deployment case studies from industry leaders
• Understanding deployment challenges and solutions

Featured Case Studies:

• Exa
• Snyk
• Weaviate
• AB InBev

Key Topics:

• Industry adoption patterns
• Production deployment strategies
• Real-world implementation challenges
• Scaling considerations for thousands of users

• Agents: Autonomous entities with specific roles, goals, and backstories
• Tasks: Discrete units of work assigned to agents
• Crews: Teams of agents working together
• Tools: Built-in and custom capabilities agents can use
• Memory: Short-term, long-term, and shared memory systems
• Guardrails: Safety and control mechanisms

• Execution Hooks: Fine-grained control over agent behavior
• Flows: Low-level orchestration control layer
• Model Context Protocol (MCP): Expanding agent tool capabilities
• Tracing: Monitoring and debugging agent decisions
• Human-in-the-Loop: Training and feedback mechanisms

• Sequential execution
• Parallel processing
• Hierarchical delegation
• Hybrid workflows
• Asynchronous operations

1. Automated Code Reviewer - Multi-agent system for code review and analysis
2. Meeting Co-Pilot - AI assistant for meeting management and follow-up
3. Deep Researcher - Comprehensive research automation system

• Basic Python programming knowledge
• Familiarity with prompt engineering concepts
• Understanding of generative AI and LLMs
• Interest in building production AI systems

# Clone the repository
git clone https://github.com/mjgrav2001/Design_Develop_Deploy_Multi_Agent_Systems_CrewAI.git

# Navigate to the repository
cd Design_Develop_Deploy_Multi_Agent_Systems_CrewAI

# Install dependencies (if requirements.txt is available)
pip install -r requirements.txt

Each module folder contains Jupyter notebooks with hands-on labs and implementations. To run the notebooks:

1. Navigate to the desired module directory
2. Launch Jupyter Notebook or JupyterLab
3. Open the notebook file (.ipynb)
4. Follow the instructions within each notebook

• Course Link: DeepLearning.AI Course Page
• Coursera: Course on Coursera
• CrewAI Documentation: Official CrewAI Docs
• CrewAI GitHub: CrewAI Repository

João Moura - Co-founder and CEO of CrewAI

João created CrewAI as an open-source framework for building and orchestrating multi-agent systems, and brings extensive experience in designing workflows for collections of agents.

• Multi AI Agent Systems with crewAI (DeepLearning.AI Short Course)
• Practical Multi AI Agents and Advanced Use Cases with crewAI (DeepLearning.AI)
• Agent Skills with Anthropic (DeepLearning.AI)

This repository contains personal course work and implementations. Feel free to:

• Fork the repository for your own learning
• Submit issues if you find errors or have questions
• Share your own implementations and improvements

This repository contains educational materials from DeepLearning.AI's course. Please respect the course's terms of use and intellectual property rights.

• DeepLearning.AI for creating and offering this comprehensive course
• João Moura and the CrewAI team for developing the framework and teaching the course
• Andrew Ng and the DeepLearning.AI team for their continued commitment to AI education

Note: This repository represents coursework completed in December 2025 as part of ongoing professional development in AI/ML engineering and multi-agent systems.