If you’re reading this, chances are you’re new to The Neural Maze.
Or maybe you’ve been subscribed for a few months but still find yourself asking: “What is this really about?”
Or maybe you’ve been supporting this publication for a long time already — in that case, thank you, amigo 🫶
No matter who you are, you’re welcome here!
This is The Neural Maze: an ecosystem built for real ML / AI engineers and AI builders.
And this article will explain everything you need to know about it 😉
Because of Substack’s format (and since a lot of you read these posts straight from your inbox), keeping things structured can be tricky.
That’s why I put this article together.
Think of it as your reference library, the place to come back to whenever you find yourself wondering:
“Where was that deep dive on Recommender Systems?”
“Did someone really train an LLM to sound like Rick Sanchez from Rick & Morty? I need to check that out.”
To keep it simple, I’ve organized everything into four main parts.
Agent Design Patterns
The Ava Series (Building a WhatsApp Agent)
The PhiloAgents Series (Building Game Simulation Agents)
The Kubrick Series (Building an MCP Agent for Video Processing)
The Phone Calling Agents Course (Building an Agent-based Call Center)
AI Project Kickstarters
Production AI Systems
AI Career Growth
Experiments and Side Projects
👨🏫 The 5 Courses
These are long-form, end-to-end courses — with multiple lessons, written deep dives, videos, and open-source code you can run yourself.
💡 Everything is available in The Neural Maze GitHub!
Course I: Agent Design Patterns
You’ll learn how to implement, from scratch, the 4 Agentic Patterns.
Course II: Ava, the WhatsApp Agent
🤝 A collaboration with Jesús Copado
You’ll learn how to build a multimodal agent connected to WhatsApp that talks and reacts like a real person. Images, voice messages, etc. Ava can handle it all.
Course III: PhiloAgents
🤝 A collaboration with Paul Iusztin
In this course, you’ll build an AI simulation that brings historical philosophers to life in an interactive game environment.
It’s split into 6 in-depth modules, where you’ll learn to design and deploy a production-ready agentic RAG system, using real LLMOps practices.
Course IV: Kubrick, the MCP Video Processing Agent
🤝 A collaboration with Alex Razvant
In this course, you’ll apply RAG to video and take it all the way to production by building an MCP Server — complete with agent monitoring, prompt versioning, and full observability.
Course V: Phone Calling Agents
🎙️ This is the first course to feature Live Session mode. The course is already underway, but you can still join the upcoming Live Sessions—don’t miss them! You’ll find all the details here:
In this course, you’ll learn how to build realtime AI voice agents using FastRTC for low-latency streaming, Superlinked for vector search, Twilio for live phone calls, and Runpod for scalable GPU deployment.
Lesson 0: The Architecture of Phone Calling Agents (Live Session 🎙️)
Lesson 1: Building Realtime Voice Agents with FastRTC (Live Session 🎙️)
Lesson 2: The Missing Layer in Modern AI Retrieval (Live Session 🎙️)
Lesson 3: How to deploy STT and TTS systems to production? (Live Session 🎙️)
Lesson 4: Deploying a multi-avatar Voice Agent with Full Tracing
Lesson 4: Deploying a multi-avatar Voice Agent with Full Tracing (Live Session 🎙️)
📚 The Neural Maze Library
Beyond the big courses, The Neural Maze is full of shorter essays, deep dives, and experiments. To keep it navigable, I’ve grouped them into a few themes:
AI Project Kickstarters
Ready-to-use templates so you can skip boilerplate and start building your AI Projects faster.
Workshops
Workshops offer a video-based, highly practical learning experience—designed to teach you by building real projects step-by-step.
Production AI Systems
Deep dive breakdowns on how production AI Systems actually work.
AI Career Growth
Sharing my own experiences, lessons, and practical advice from the path of becoming an ML / AI Engineer.
Experiments and Side Projects
Not everything I build is meant for production. Some projects are just for fun, curiosity, or testing ideas that might fail spectacularly …. and that’s the point.
💬 The Neural Chat
This is our little corner for connection. A space where subscribers can meet each other, ask questions, and swap ideas on all things ML / AI.
Whether you’re here to learn, share, or just explore … come say hello and join the conversation!
⭐ Why Go Premium?
Premium gives you full access to The Neural Maze — everything you need to build real AI systems, end to end.
With Premium, you get:
Full access to 70+ articles, including 5 complete end-to-end courses
Project kickstarter templates to accelerate your builds
Access to live sessions with coding demos and interactive Q&A
💡 Best value: the Annual plan is 61% cheaper than subscribing monthly.
❓ So, where should you start?
You’ve got a lot of resources to explore, and it can feel like a lot at first. That’s why I’ve laid out a simple path to help you move forward with confidence.
1️⃣ Start with the Courses
If you’re serious about leveling up, begin with the 5-course path. Each one builds on the last, guiding you through new technologies, techniques, and architectures:
🏁 → Agent Design Course → Ava → PhiloAgents → Kubrick → Phone Calling Agents → 🏁
Finish all five, and you won’t just be better … you’ll be in a whole new place.
2️⃣ Want some advice?
Not in the mood to code? No problem.
Head over to the AI Career Advice section. It’s packed with personal lessons, honest takes, and insights to keep you motivated.
3️⃣ Keep Exploring
Want more? Keep reading—here are some of my bestselling articles that a lot of people have found helpful:
So … what’s your next step? 😉
Thank you for sharing this, had started using LiteLLM very recently but this made it much cleaner approach for learning.