Now What?
You finished the book. Here is what to do next.
Now What?
You made it. The whole book. From x = 5 to deploying a GPT on the internet.
Let's look at what you actually know now:
The Journey
Part 1 — You learned Python. Variables, loops, functions. The language that runs the AI world.
Part 2 — You met 13 ML algorithms. Each one a character:
- The Line taught you prediction
- The Sorting Hat taught you classification
- The Committee taught you neural networks
- The 20 Questions taught you decisions
- The Mob taught you ensemble wisdom
- The Mean Girls taught you clustering
- The Blind Hiker taught you optimization
- The Memorizer taught you what not to do
- The Eyeball taught you to see
- The Goldfish taught you to remember
- The Attention Seeker taught you to focus
- The Dog Trainer taught you rewards
- The Forger taught you creation
Part 3 — You built real things:
- Your own backpropagation engine (the secret sauce)
- Your own GPT (the architecture behind ChatGPT)
- An image classifier (computer vision)
- A text generator (fine-tuning)
- A deployed web app (real users, real internet)
What To Build Next
The best way to learn is to build something you care about. Here are ideas:
| Project | What You Need | Difficulty |
|---|---|---|
| Spam detector | Text classification (ML 2) | Easy |
| Music recommender | Clustering (ML 6) | Easy |
| Stock price predictor | Regression (ML 1) + RNNs (ML 10) | Medium |
| Face detector | CNNs (ML 9) + PyTorch (App 3) | Medium |
| Your own ChatGPT | Transformers (ML 11) + Fine-tuning (App 4) | Hard |
| AI game player | Reinforcement Learning (ML 12) | Hard |
| Image generator | GANs (ML 13) + PyTorch | Hard |
Where To Go Deeper
If you want more math:
- 3Blue1Brown's "Neural Networks" series on YouTube
- Stanford CS229 (Andrew Ng's Machine Learning course)
If you want more code:
- Andrej Karpathy's Neural Networks: Zero to Hero playlist
- fast.ai's "Practical Deep Learning for Coders"
If you want to build products:
- HuggingFace's model hub — thousands of pre-trained models you can use today
- LangChain / LlamaIndex for building AI apps with LLMs
If you want to go bleeding edge:
- Read papers on arXiv (start with "Attention Is All You Need")
- Follow AI researchers on X/Twitter
- Contribute to open-source AI projects
One Last Thing
You started this book as someone who "did not understand Neural Networks at parties." Now you can build one from scratch, train it, and deploy it on the internet.
You are still an idiot. But now you are a dangerous idiot.
Go build something dumb. Then make it smart.