Before I give advice, do you want to focus on Python coding or high-level logic? Mimo: Learn Coding/Programming is a decent option for your commute depending on your goals.

Here's what I recommend:

1. DataCamp Mobile App: Seriously the best for ML fundamentals. Wait for their big sales—I’ve seen annual subs drop to $149.
2. Enki App: Highkey underrated for AI. It’s gamified and basically perfect for a 30-min commute. Pro is only like $8/month too.

I'd be careful with free versions tho, technical depth might be lacking unless u pay. gl!

Coming back to this—tbh I’ve spent years testing apps on my commute. Here’s my 2024 shortlist:

- Brilliant.org: Still the GOAT for neural network logic.
- Codecademy Go: Great for practicing syntax on the move.
- Udacity: Best for technical depth and current specs.

Honestly, don't cheap out—the paid curricula stay way more up-to-date. It's worth the investment. GL with the commute!!

👆 this

Honestly, looking at the 2024 landscape, you gotta check out Coursera if you want the actual industry-standard specs. Their partnership with DeepLearning.AI is basically the benchmark for neural networks - it's kinda more "academic" but their mobile app is SO good for downloading high-level lectures for when the train signal drops. If you compare that to Udemy, the market vibe is totally different. Udemy is way better for "market-fresh" tutorials on things like RAG or fine-tuning specific LLMs that just launched, though the technical depth is a bit hit-or-miss? I think Coursera is better for the fundamental "why" while Udemy wins on the "how to do it right NOW" side of things. Both are solid for a commute, but Coursera’s curriculum stays a bit more authoritative for the 2024 tech stack, I think.

Just saw this thread and honestly BlackpoolTowerView is spot on. It is so frustrating how every company is trying to cash in on the AI hype right now without actually providing deep value. Honestly its ridiculous how much these platforms are charging for what usually ends up being outdated or surface-level content. It drives me crazy that you have to shell out a fortune every month just to get past a paywall, only to find out the curriculum is basically just fluff. These companies dont seem to care about actual performance or technical reliability anymore, they just want to hook you with a flashy UI. It feels like a total scam sometimes when you realize youre basically paying for a pretty interface around info that isnt even vetted. The lack of focus on real metrics and model efficiency in these courses is just sad. Its such a mess trying to find anything legit these days without getting drained by subscription fatigue. Everything is just so overpriced and under-delivered lately.

Honestly, I’ve spent way too much money on subs before realizing that the DIY route is where the real growth happens if you’re already tech-adjacent. I started my journey just using Google Colab on my phone and iPad during my commute. It’s basically a free Jupyter notebook environment in the cloud. Instead of following a rigid app curriculum, I just started cloning repos from GitHub and trying to get inference running on smaller models. It's wierd how much more you learn when you have to fix your own broken dependencies. Quick tip: Skip the hand-holding apps if you want to understand the actual architecture. Focus on getting comfortable with Kaggle and their mobile-friendly notebooks. Understanding how to pipe data into a transformer model yourself is worth ten "gamified" streaks, right?

Yeah, I'm kind of in the same boat as a total beginner. I’ve been trying to stay consistent for about six months now and honestly, it’s a struggle! I started with the fun gamified stuff but I found myself just memorizing patterns instead of actually *learning* the logic. Well actually, I realized I needed more real-world context to keep me interested during my commute. Here are a few things I’ve been sticking with lately that are pretty budget-friendly:
* Kaggle - The mobile web version is surprisingly good for their free micro-courses. It feels more 'grown-up' than some apps because you're looking at real data.
* LinkedIn Learning - I got this free through my local library! Their 2024 AI tracks are really polished and they have a great offline mode for when the train hits a dead zone.
* YouTube - I basically just download long-form explainers from channels like StatQuest. It's not 'interactive' but it makes the theory stick better for me. I don't know if I'm doing it right yet, but focusing on these has definitely helped it feel less like a chore and more like a hobby, you know? Good luck!

Honestly I think everyone focuses too much on the interface and not enough on the actual performance metrics of the models you are learning about. My journey really took off when I stopped looking at apps as just a series of lessons and started treating them as a lab for benchmarking. I spent most of last year obsessing over things like tokens per second and how different architectures handle context windows because thats where the rubber meets the road in 2024. I found that my current setup for learning involves a lot more raw testing than just watching videos because you dont really get a feel for the tech until you see it struggle with a complex prompt.

• tracking inference latency on different hardware configurations
• comparing quantization levels to see how they affect accuracy
• monitoring vram usage while running local experiments Tbh if an app doesnt let you see the raw output or doesnt explain the performance trade-offs of the models its teaching you its basically just a toy. I learned way more by breaking things and measuring the fallout than I ever did by just getting a gold star in a gamified quiz. Focus on the data and the specs and the rest will start making sense way faster.

Noted!