Been on this platform for about 4 years now and drafted many articles I never published...Guess I finally have a reason to post one (expect more after this lol), maybe a few reels and youtube videos at this rate😂.
If you've been anywhere near tech over the last couple of years, you'll notice everyone tryna get on newer techs and frameworks and for sure, you've probably felt it too.
You blink once, and suddenly there's a new LLM everyone is talking about. Blink again, and someone has launched "the next ChatGPT" or another funny chatgpt wrapper. Open LinkedIn for five minutes and there's another startup claiming they've built an AI agent that will replace... well, everything, like being on LinkedIn ain't oppressing enough🤦🏽♂️.
Ngl, keeping up with AI these days feels like trying to drink from a firehose. And the funny part? I already work in AI. These days I spend my weekdays working as a Machine Learning Engineer at a space research agency. Before that, I spent years building frontend and backend applications, prototyping robots and autonomous systems as academically I came through Electrical & Electronics Engineering before later studying Computer Engineering. Looking back, each stage has shaped how I approach AI today. Even then, I still find myself thinking, "Wait... should I be learning this too?"
That's probably one of the biggest lessons I've learned the last two years, that working in AI doesn't mean you stop learning. If anything, it means you have even more reasons to keep learning.
So when a friend introduced me to Scrimba's AI Engineer Path a few months ago, I wasn't necessarily looking for another course. I was looking for a better way to learn.
There's No Shortage of AI Resources...
If you've tried learning AI recently, you know the struggle. There are YouTube playlists, official documentation, GitHub repositories, research papers, blogs, Discord communities, X (formerly Twitter), where every weekend feels like an AI hackathon. And then there are the courses.
I've taken quite a few over the years covering machine learning, deep learning, MLOps, and more recently, AI engineering. Tbh, some were fantastic. Others... well, let's just say I collected more certificates than practical projects yktv😭. Tutorial hopping should honestly count as cardio at this point.
My own background has shaped how I approach learning. Academically, I come from Electrical & Electronics Engineering (undergrad) and Computer Engineering (Grad). Before moving deeper into AI, I spent years building frontend and backend applications. That software engineering foundation has probably helped me more than I realized.
When you already think in terms of APIs, databases, authentication, deployment, and user experience, building AI applications starts feeling less like magic and more like adding another capability to software you've been building all along.
Still, even with that background, I found myself overwhelmed by just how fast everything was moving( I promise you, as I write this rn, just saw a reel of GLM5.2 being released lmao). Anyhoo, back to story. Began trying AI engineering courses, then a colleague suggested I try Scrimba. His reasoning actually made a lot of sense. He knew I was juggling a full-time role at the space research agency while also spending evenings and weekends building my startup (look out, it's my biggest motivation for trying to start publishing). Sitting through hours of passive video courses wasn't exactly realistic anymore.
Instead of just recommending the platform, he offered me access to his account for a few weeks so I could see what made it different before deciding whether it was worth committing to myself. Fair enough. I figured I'd give it a shot.
The First Thing That Felt Different
One thing I noticed almost immediately was that I wasn't just watching someone code. I was expected to participate. If you've never used Scrimba before, the lessons are interactive. Instead of pausing a video, opening VS Code, recreating the project, realizing you've missed three tiny steps, then rewinding the lesson for the fourth time (we've all been there 😅), you're editing the instructor's code directly inside the lesson itself.
It sounds like a small difference. It really isn't. There were several moments where I'd pause halfway through a lesson, not because I was confused, but because I wanted to see what would happen if I changed something.
Sometimes I'd tweak a prompt, other times I'd modify the logic. Occasionally, I'd break everything. Which, honestly, is usually when the learning starts.
That interactive workflow kept me engaged in a way passive videos rarely do. Instead of thinking, "I'll remember this later," I was actually using the ideas immediately.
And unsurprisingly...I remembered them better.
Halfway Through the AI Engineer Path
I'm currently about halfway through the AI Engineer Path, so I definitely can't speak for the entire curriculum yet. But even at this stage, it's already shifted how I think about AI.
Coming from machine learning, I was naturally used to thinking about datasets, training models, evaluation metrics, optimization, and deployment.
Those skills are still valuable. But AI engineering introduces another perspective. Instead of asking, "How do I build the model?". You're often asking, "How do I build a useful product around existing models?"
That distinction sounds subtle but it's actually huge. So far, I've spent time reinforcing concepts around:
- Working with Large Language Models through APIs rather than training foundation models from scratch.
- Prompt engineering and how small prompt changes can completely change outputs.
- Designing AI workflows instead of treating LLMs like magic black boxes.
- Thinking about retrieval-augmented generation (RAG) conceptually and why external knowledge matters.
- Understanding embeddings at a higher level and where they fit into modern applications.
- Getting early exposure to AI agents and what "agentic" systems actually mean in practice.
Interestingly, my background in machine learning helped me understand some of these ideas more quickly. Knowing how models are built makes it easier to appreciate their strengths and limitations.
But what surprised me was how much I still had to learn. Building intelligent software isn't exactly the same as building machine learning models.
There's a whole layer of application design, orchestration, user interaction, prompt iteration, and system thinking that becomes incredibly important.
That shift in mindset has probably been the biggest takeaway so far.
AI Isn't Slowing Down Anytime Soon
If anything has become clear over the past year, it's that AI moves ridiculously fast.
Every week there's another framework.
Another SDK.
Another agent library.
Another benchmark.
Another model.
I swear every time I open LinkedIn someone has built the next AI startup.
By the time I finish reading about one model, three newer ones have already dropped😮💨.
Sometimes it feels impossible to keep up. But I've slowly realized something. Maybe the goal isn't to chase every new release. Maybe it's to build a strong enough foundation that learning the next tool becomes easier.
Frameworks will change.
Model names will change.
Pricing will change.
But understanding how AI systems fit into real software...that's surely a skill that keeps paying off (at least till AI will finally take your job not mine 😂😭).
Would I Recommend Scrimba?
Yeah, I would. Especially if you're already comfortable writing code and want to start building AI-powered applications instead of only learning the theory.
For me, the biggest advantage wasn't that Scrimba magically taught concepts nobody else does. It was how those concepts were taught. The interactive approach kept me engaged.
It encouraged experimentation. It made it easier to connect lessons to things I encounter at work.
And perhaps most importantly...It made me look forward to opening the next lesson.
That said, I don't think any single learning platform is enough.
Not Scrimba. Not YouTube. Not university. Not documentation.
The best learning happens when you combine multiple resources.
For me, that usually means pairing structured courses with official documentation, GitHub projects, personal experiments, YouTube deep dives, research papers when I need more depth, and conversations with other developers.
Learning AI today isn't about finding the perfect course, cos honestly them things don't exist.
It's about staying curious long enough to keep building.
A Small Thank You
To the team behind Scrimba, thank you. The learn-by-doing approach genuinely made a difference for me.
You've managed to take topics that can feel intimidating at first and present them in a way that encourages experimentation instead of making mistakes feel expensive.
As someone who enjoys building things, that's probably what I've appreciated the most. I'm only halfway through the AI Engineer Path, so there's still plenty left for me to learn.
And honestly...
I'm looking forward to it.
Because if there's one thing AI has taught me, it's this:
No one is ever truly "caught up." The field moves too fast for that.
The real advantage isn't knowing every framework or every new model the day it's released. It's staying curious...
Building consistently.
Breaking things.
Fixing them.
Then repeating the process.
AI isn't slowing down anytime soon🧠. So we might as well enjoy the ride.
Thanks for reading if you made this far. I hope I didn't bore you
If You're Thinking About Learning AI Engineering
If you're a developer who's comfortable writing code and wants to start building AI-powered applications, I'd genuinely recommend checking out Scrimba's AI Engineer Path.
What I appreciate most isn't that it promises to teach some secret AI formula, but that it encourages you to learn by building. That style just clicked with me.
If you're curious and want to explore the course yourself, you can check it out here:
👉 https://scrimba.com/?via=community
Top comments (0)