Since I started The Neural Maze, my goal has been to build a place for real, practical AI content — the kind that challenges you, makes you think, and ultimately helps you grow into a high-skilled ML / AI engineer ready for top roles.
Today, I’m sharing four new hands-on projects from the past few months to help you level up through our core philosophy — learning by building.
You can also check out my previous list of projects for a total of 10 hands-on builds to sharpen your skills.
Substack Search Engine
💻 Want to explore the code? Get the full repository here
Let’s begin with this masterpiece by Benito Martin . Honestly, since he started his newsletter, AI Echoes, he’s been putting out top-quality content nonstop.
It was my immense pleasure and privilege to talk about his Substack Search Engine course here on The Neural Maze.
In a nutshell, this course is one of the best real-world examples of what it means to build a production-grade RAG system. And the best part? It’s so useful.
Imagine being able to search ANY newsletter content from your favorite Substack creators. Well… that’s exactly what you’ll get out of this course.
It includes 4 deep-dive lessons, an amazing repo, and a full video walkthrough to help you get started.
It also includes a video walkthrough, made by myself, that will show you how the repo works and how to get started with this project. Check it out here:
Production-ready ML Application
💻 Want to explore the code? Get the full repository here
In this course, Anas Riad will teach you how to build a production-ready ML system using a powerful stack of modern tools and libraries. Finally, you’ll step out of the Jupyter Notebook and see how real ML systems are built!
You’ll learn how to dockerize your project, expose your model with FastAPI, use tools like MLflow and Optuna, and—best of all—deploy everything to AWS.
You can find the full deep dive in this Substack article that Anas wrote for The Neural Maze:
And the best part? It also has a video version! So if you’re more of a video person, you’re in luck…
AWS Agentic Architectures
💻 Want to explore the code? Get the full repository here
This project feels really special to me — and not just because it’s super good, but because:
It’s based on one of my first courses for The Neural Maze
And it was created by one of my premium subscribers!
Ladies and gentlemen, let me introduce you to Marcelo Acosta Cavalero , the AWS wizard (and don’t forget to subscribe to his own newsletter, Build With AWS).
In this course, Marcelo will teach us how to build a WhatsApp Multimodal Agent entirely using AWS technologies (just check the diagram above): Lambda functions, Bedrock Agents, Bedrock AgentCore, and more.
A true gem for those of you who want to see what real agents look like in AWS.
Phone Calling Agents
💻 Want to explore the code? Get the full repository here
And last, my newest course with the one and only Jesús Copado.
We’re only on Lesson 2 so far, so the journey has just begun, but this course comes with something totally new: live sessions.
If you’re a premium subscriber, you can hop in with Jesús and me every Sunday for a 1h 30m live call where we break down the week’s lessons, answer your questions, and brainstorm your ideas and improvements together.
🎙️ You can explore all the lessons released so far, plus the recordings of past live sessions, right here!
You can also check the upcoming Live Session here:
That’s all for this article … for now. Give it a few months and I’m sure we’ll have enough new courses to make a whole new list! 🤣
Wishing you a wonderful Saturday and, for my premium subscribers…
See you tomorrow! 👋
Thanks for these resources
Thank you it very useful , I want to be expert in Agentic AI and now I learn both theory and techniques but I am confuse about….
When can I start to build my own Agentic AI ideas?
Do you think that applications from other people's projects, like the ones you are presenting here, are useful for learning?