Hello Reader,

Hope you're having a great start to the week!

This past week, I’ve been diving into DeepSeek-R1 and exploring some AI books, and I wanted to share some thoughts and insights with you.

Building AI Applications

A few weeks ago, DeepSeek made headlines as the most powerful open-source reasoning model. I put it to the test in a video, comparing its performance against OpenAI's GPT-4o. Spoiler: It's impressive—but not without flaws!

But more importantly, DeepSeek represents something bigger: open-source models are making AI development easier and more affordable than ever. Whether you're a solo developer or part of a large company, you can now build powerful AI applications at little to no cost—especially if you know how to code.

Andrew Ng has pointed out that while AI foundation models get most of the attention, most of the opportunities lie in building applications on top of them.

That said, it might need to take his words with a grain of salt. Generative AI is powerful, but depending on the use cases, it's not always the best solution. If a simple regression model or a traditional algorithm does the job just fine, there's no need to add the complexity and cost of an LLM.

Unfortunately, fear of missing out (FOMO) often drives companies to implement AI where it’s unnecessary. The LinkedIn post below makes this point real.

So the real opportunity isn’t just in using AI, but perhaps also in knowing when not to use it..

The Rise of AI Engineering

Lately, I’ve been reading "AI Engineering: Building Applications with Foundation Models" by Chip Huyen. One of the key ideas in the book is how AI engineering is emerging as a distinct discipline in tech.

Anyone can put together a basic AI app, but the real challenge is building robust, reliable AI applications that solve real-world problems. That’s where AI engineers come in, and this role seems to be growing fast.

Looking at recent trends, open-source AI engineering frameworks have been growing faster than established software tools like Vue, React, and even Bitcoin over the past few years.

If you're familiar with machine learning and data science, you've probably heard of ML engineering. Traditionally, ML engineers focus on developing new models for specific tasks.

AI engineering, on the other hand, is more about leveraging existing models and adapting them to different applications. While there's some overlap, the focus shifts from model creation to model application.

👉 The AI Engineering Stack

In AI Engineering, Chip Huyen outlines 3 key layers of the AI engineering stack:

1. Application Development – Just like using GPT-4o to build your own AI app. This involves crafting effective prompts, incorporating relevant context (like a knowledge base), and designing a solid user interface.
2. Model Development – Fine-tuning models, optimizing inference, and handling data collection and engineering when needed.
3. Infrastructure – Managing cloud resources, compute power, monitoring, and maintaining AI applications at scale.

To see how AI engineering is reflected in the job market, I’m analyzing over 3 million data-related job postings. My goal is to identify hiring trends and confirm whether AI engineering is becoming more popular. I’ll share my findings with you in one of my coming YouTube videos!

‼️🧭 By the way, there’s still a $50 New Year discount on my Python for AI Projects course until February 14—don’t miss out! It’s packed with everything you need to stay ahead in AI and transform your skills in 2025.

Click below for more details 👇.

Have a great week ahead! 😊

Thu