Intermediate course
AI Engineer Agentic Track: The Complete Agent & MCP Course
Master AI Agents in 30 days: build 8 real-world projects with OpenAI Agents SDK, CrewAI, LangGraph, AutoGen and MCP.
Intermediate
Course facts
- Last updated 02/2026
- English English [Auto], Arabic [Auto] , 24 more
- Instructor: Ed Donner, Ligency
- AI agents, automation and connected workflows
What you'll learn
Practical outcomes
- Project 1: Career Digital Twin. Build and deploy your own Agent to represent you to potential future employers.
- Project 2: SDR Agent. An instant business application: create Sales Representatives that craft and send professional emails .
- Project 3: Deep Research. Make your own version of the essential Agentic use case: a team of Agents that carry out extensive research on any topic you choose.
- Project 4: Build a Stock Picker Agent in minutes with CrewAI—automate your search for investment gems!
- Project 5: Deploy your own 4-Agent Engineering Team—manage, build, and test software apps with CrewAI and Coder Agents in Docker!
- Project 6: Build your own version of OpenAI’s Operator Agent—your Sidekick works with you inside your browser via LangGraph!
- Project 7: Agent Creator—an Agent that builds and launches new Agents using AutoGen, unlocking endless AI possibilities!
- Project 8: Capstone—build a Trading Floor with 4 Agents making autonomous trades, powered by 6 MCP servers and 44 tools!
Curriculum
6 sections • 130 lectures • 17h 2m total length
Week 127 lectures • 4hr 6min
- Day 1 - Autonomous AI Agent Demo: Using N8n to Control Smart Home Devices07:15
- Day 1 - AI Agent Frameworks Explained: OpenAI SDK, Crew AI, LangGraph & AutoGen11:36
- Day 1 - Agent Engineering Setup: Understanding Cursor IDE, UV & API Options11:50
- Day 1 - Windows Setup for AI Development: Git, Cursor IDE & UV Package Manager20:54
- Day 1 - Setting Up Your Mac for AI Projects: GitHub, Cursor IDE & OpenAI API Key19:50
- Day 1 - Building Your First Agentic AI Workflow with OpenAI API: Step-by-Step17:35
- Day 1 - Introduction to Agentic AI: Creating Multi-Step LLM Workflows + Autonomy01:34
- Day 2 - Building Effective Agents: LLM Autonomy & Tool Integration Explained06:13
- Day 2 - 5 Essential LLM Workflow Design Patterns for Building Robust AI Systems08:32
- Day 2 - Understanding Agent vs Workflow Patterns in LLM Application Design06:39
- Day 3 - Orchestrating Multiple LLMs: Comparing GPT-4o, Claude, Gemini & DeepSeek10:16
- Day 3 - Multi-LLM API Integration: Comparing OpenAI, Anthropic & Other Models09:47
- Day 3 - Comparing LLM APIs: Using OpenAI Client Library with Claude, Gemini & ++12:56
- Day 3 - Multi-Model Orchestration: Creating a System to Evaluate AI Responses10:52
- Day 3 - Connecting Agentic Patterns to Tool Use: Essential AI Building Blocks00:35
- Day 4 - Comparing AI Agent Frameworks: Simplicity vs Power in LLM Orchestration06:30
- Day 4 - Resources vs. Tools: Two Ways to Enhance LLM Capabilities in Agentic AI07:45
- Day 4 - Build a Web Chatbot That Acts Like You Using Gradio & OpenAI09:48
- Day 4 - Using Gemini to Evaluate GPT-4 Responses: A Multi-LLM Pipeline13:15
- Day 4 - Building Agentic LLM Workflows: Resources, Tools & Structured Outputs01:22
- Day 5 - Building Your Career Alter Ego: LLM Function Calling with Push Alerts08:20
- Day 5 - LLM Tool Calls Demystified: How to Process and Execute Function Requests05:44
- Day 5- Building AI Assistants: Implementing Tools for Handling Unknown Questions02:44
- Day 5 - Creating & Deploying an AI Agent: From Chat Loop to HuggingFace Spaces10:44
- Day 5 - Deploying Career Conversation Chatbots to Gradio08:44
- Day 5 - Foundation Week Wrap-up: Building Complete AI Agents with APIs & Tools01:35
- Day 5 [Extra] - Building Your First Agent Loop with OpenAI Tools from Scratch13:23
Week 221 lectures • 2hr 26min
- Day 1 - Understanding Async Python: The Foundation for OpenAI Agents SDK11:44
- Day 1 - OpenAI Agents SDK Fundamentals: Creating, Tracing, and Running Agents05:15
- Day 1 - Introduction to Agent, Runner, and Trace Classes in OpenAI Agents SDK08:36
- Day 1 - Vibe Coding: 5 Essential Tips for Efficient Code Generation with LLMs06:55
- Day 1 - OpenAI Agents SDK: Understanding Core Concepts for AI Development00:16
- Day 2 - Build AI Sales Agents with SendGrid: Tools & Collaboration in Agent SDK07:26
- Day 2 - Concurrent LLM Calls: Implementing Asyncio for Parallel Agent Execution09:00
- Day 2 - Converting Agents into Tools: Building Hierarchical AI Systems06:27
- Day 2 - Agent Control Flow: When to Use Handoffs vs. Agents as Tools07:47
- Day 2 - From Function Calls to Agent Autonomy: Sales Automation with OpenAI SDK06:33
- Day 2 - Agentic AI for Business: Creating Interactive Sales Outreach Tools01:01
- Day 3- Multi-Model Integration: Using Gemini, DeepSeek & Groq with OpenAI Agents07:46
- Day 3 - Implementing Guardrails & Structured Outputs for Robust AI Agent Systems10:11
- Day 3- AI Safety in Practice: Implementing Guardrails for LLM Agent Applications05:25
- Day 4 - Building Deep Research Agents: Implementing OpenAI's Web Search Tool08:49
- Day 4 - Building a Planner Agent: Using Structured Outputs with Pydantic in AI07:58
- Day 4 - Building an End-to-End Research Pipeline with GPT-4 Agents & Async Tasks10:28
- Day 4 - Building a Deep Research Agent: Parallel Searches with AsyncIO03:55
- Day 5 - Building a Modular AI Research System with Gradio UI Implementation12:26
- Day 5 - Deep Research App: Gradio to Visualize & Monitor Autonomous AI Agents03:36
- Day 5 - Deploying Smart Research Agents with Gradio and HuggingFace Spaces03:59
Week 319 lectures • 2hr 32min
- Day 1 - Crew AI Framework: Creating Collaborative AI Agent Teams06:02
- Day 1 - Crew AI Framework Explained: Agents, Tasks & Processing Modes Tutorial07:44
- Day 1 - Crew AI & LightLLM: Flexible Framework for Integrating Multiple LLMs04:46
- Day 1 - Crew AI Tutorial: Setting Up a Debate Project with GPT-4o mini09:20
- Day 1 - How to Create an AI Debate System Using Crew AI and Multiple LLMs12:20
- Day 1 - Building AI Debate Systems with CrewAI: Compare Different LLMs02:07
- Day 2 - Building Crew AI Projects: Tools, Context & Google Search Integration06:03
- Day 2 - Building Multi-Agent Financial Research Systems with Crew.ai10:46
- Day 2- Enhancing AI Agents with Web Search: Solving the Knowledge Cutoff Problem05:44
- Day 3 - Building a Crew AI Stock Picker: Multi-Agent System for Investments07:10
- Day 3 - Implementing Pydantic Outputs in Crew AI: Stock Picker Agent Tutorial08:39
- Day 3 - Custom Tool Development for Crew AI: JSON Schema & Push Notifications08:49
- Day 4 - Crew AI Memory: Vector Storage & SQL Implementation for AI Agents12:14
- Day 4 - Crew AI for Coding Tasks: Agents That Generate & Run Python Code08:06
- Day 4 - Create a Python-Writing AI Agent: Practical Implementation with Crew AI06:04
- Day 5 - Building AI Teams: Configure Crew AI for Collaborative Development10:21
- Day 5 - Collaborative AI Agent Development for a Stock Trading Framework08:09
- Day 5 - Building a Trading Application Using GPT-4o & Claude08:30
- Day 5 - From Single Modules to Complete Systems: Advanced CrewAI Techniques08:49
Week 423 lectures • 2hr 59min
- Day 1 - LangGraph Explained: Graph-Based Architecture for Robust AI Agents10:12
- Day 1 - LangGraph Explained: Framework, Studio, and Platform Components Compared05:43
- Day 1 - LangGraph Theory: Core Components for Building Advanced Agent Systems09:52
- Day 2 - LangGraph Deep Dive: Managing State in Graph-Based Agent Workflows05:47
- Day 2 - Mastering LangGraph: How to Define State Objects & Use Reducers07:12
- Day 2 - LangGraph Fundamentals: Creating Nodes, Edges & Workflows Step-by-Step06:22
- Day 2 - LangGraph Tutorial: Building an OpenAI Chatbot with Graph Structures03:59
- Day 3 - LangGraph Advanced Tutorial: Super Steps & Checkpointing Explained05:47
- Day 3 - Setting Up Langsmith & Creating Custom Tools for LangGraph Applications06:31
- Day 3 - LangGraph Tool Calling: Working with Conditional Edges & Tool Nodes11:31
- Day 3 - LangGraph Checkpointing: How to Maintain Memory Between Conversations08:56
- Day 3 - Building Persistent AI Memory with SQLite: LangGraph State Management05:57
- Day 4 - Playwright Integration with LangGraph: Creating Web-Browsing AI Agents08:41
- Day 4 - Create AI Web Assistants: Playwright, LangChain & Gradio Implementation08:03
- Day 4 - LLM Evaluator Agents: Creating Feedback Loops with Structured Outputs09:16
- Day 4- Creating LLM Feedback Loops: Worker-Evaluator Implementation in LangGraph10:28
- Day 4 - Building an AI Sidekick Using LangGraph, Gradio & Browser Automation08:49
- Day 5 - Agentic AI: Add Web Search, File System & Python REPL to Your Assistant05:34
- Day 5 - LangChain Tool Integration: Building a Powerful AI Sidekick from Scratch09:59
- Day 5 - Creating AI Workflows: Graph Builders & Node Communication Techniques08:33
- Day 5 - Creating Isolated User Sessions in Gradio Apps Using State Management06:00
- Day 5 - Inside AI Feedback Loops: Seeing How AI Evaluates & Corrects Errors12:15
- Day 5 - AI Assistant Upgrades: Memory, Clarifying Questions & Custom Tools03:46
Week 517 lectures • 2hr 10min
- Day 1 - Microsoft Autogen 0.5.1: AI Agent Framework Explained for Beginners07:36
- Day 1 - AutoGen vs Other Agent Frameworks: Features & Components Compared05:48
- Day 1 - AutoGen Agent Chat Tutorial: Creating Tools and Database Integration09:44
- Day 1 - Essential AI Components: Models, Messages & Agents Explained00:14
- Day 2 - Advanced Autogen Agent Chat: Multimodal Features & Structured Outputs09:27
- Day 2 - Implementing Primary and Evaluator Agents in AutoGen with Langchain13:36
- Day 2 - Headless Web Scraping Tutorial: MCP Server Fetch Integration in AutoGen08:12
- Day 3 - AutoGen Core: The Backbone of Distributed Agent Communications05:01
- Day 3 - Agent Communication in Autogen Core: Message Handlers & Dispatching08:43
- Day 3 - AutoGenCore Agent Registration and Message Handling: Practical Examples09:04
- Day 3 - AutoGenCore Standalone Agents: Rock Paper Scissors with GPT-4o & Llama07:28
- Day 4 - Autogen Core Distributed Runtime: Architecture & Components Explained03:00
- Day 4 - Implementing Distributed AI Agents with AutoGen Core and gRPC Runtime10:28
- Day 4 - Building Distributed Agent Systems: AutoGen Cross-Process Communication04:21
- Day 5 - Creating Autonomous Agents That Write & Deploy Other Agents in AutoGen05:03
- Day 5 - Implementing Agent-to-Agent Messaging with Autogen Core & Templates10:46
- Day 5 - Creating Autonomous AI Agents that Collaborate Using Async Python11:44
Week 6 - MCP23 lectures • 2hr 50min
- Day 1 - Intro to MCP: The USB-C of Agentic AI06:34
- Day 1 - Understanding MCP Hosts, Clients, and Servers09:18
- Day 1 - Using MCP Servers with OpenAI Agents SDK08:06
- Day 1 - Exploring Node-Based MCP Servers & Tool Access05:01
- Day 1 - Building an Agent That Uses Multiple MCP Servers11:14
- Day 1 - MCP Marketplaces & Security Considerations02:51
- Day 2 - Intro to Week 6 Day 2: Building Your Own MCP Server04:43
- Day 2 - Wiring Business Logic into Your MCP Server06:29
- Day 2 - Creating Client Code to Use Your MCP Server12:05
- Day 2 - Wrap-Up: Capabilities of Your Custom MCP Server00:20
- Day 3 - Exploring Types of MCP Servers and Agent Memory08:19
- Day 3 - Brave Search API: MCP Server Calling the Web08:30
- Day 3 - Integrating Polygon API for Stock Market Data05:26
- Day 3 - Advanced Market Tools Using Paid Polygon Plan05:52
- Day 4 - What’s Next: Launching Our Agent Trading Floor07:45
- Day 4 - Viewing the User Interface for Trading Activity10:45
- Day 4 - How Trading Agents Operate and Make Decisions07:07
- Day 4 - Portfolio Management with Four Autonomous Agents10:20
- Day 5 - Which Agent Framework Should You Pick?09:10
- Day 5 - Key Settings and Launching the Trading System05:51
- Day 5 - Advice for Selecting Agentic Frameworks08:27
- Day 5 - 10 Essential Lessons for Building Agent Solutions08:15
- Day 5 - Course Recap and Final Goodbye – Keep Building!07:14
Who it is for
- Well, perhaps I’m biased, but I’d say: anyone and everyone! If you’re fascinated in the potential of Agents and hungry to have the skills to create powerful Agentic AI – then you’ve come to the right place. While it’s most suited to those with programming experience, I’ve designed the course to work for all backgrounds.
Course description
Overview
2026 is nothing short of a watershed moment for AI Agents. It has never been more important to be an expert with Agentic AI. And that is precisely the goal of this course: to equip you with the skills and expertise to design, build and deploy Autonomous AI Agents, opening up new career and commercial opportunities. This is an intensive 6-week program to master Agentic AI. We start by building foundational expertise, connecting LLMs using proven design patterns. Then, each week, we upskill with new frameworks: OpenAI Agents SDK, CrewAI, LangGraph and Autogen. The course culminates with a full week on the remarkable opportunities opened up by MCP. Above all, this is a hands-on course. I’m a big believer that the best way to learn is by DOING. So please prepare to roll up your sleeves! We’ll build 8 real-world projects; some are astonishing, some are intriguing, and some are quite surreal. But one thing’s for sure: all are powerful demonstrations of Agentic AI’s potential to utterly transform the business landscape. So come join me on this comprehensive 6-week journey. By the end, you will have mastered Agentic AI. You will have expertise in all the major frameworks. You’ll be well-versed in the strengths and traps of Agentic AI. You’ll confidently unleash Autonomous Agents to solve real-world commercial problems. And along the way, you’ll have had a whole lot of fun with the astounding, groundbreaking technology that is Agentic AI.
Instructor
Ed Donner, Ligency
Ed Donner AI startup co-founder and leader; Gen AI and LLM instructor Ed Donner is a technology leader and repeat founder of AI startups. He’s the co-founder and CTO of Nebula, the platform to source, understand, engage and manage talent, using Generative AI and other forms of machine learning. Nebula’s long-term goal is to help people discover their potential and pursue their reason for being. Previously, Ed was the founder and CEO of AI startup untapt, an Accenture Fintech Innovation Lab company, acquired in 2021. Before that, Ed was a Managing Director at JPMorgan Chase, leading a team of 300 software engineers in Risk Technology across 3 continents, after a 15-year technology career on Wall Street. Ed holds a patent for a Deep Learning matching engine issued in 2023, and an MA in Physics from Oxford.