If you’re learning AWS for machine learning, there are really only six AWS skills you need to learn first.

The problem is, AWS has HUNDREDS of services to choose from, so it’s really easy to waste time learning stuff you won’t lose, or just get overwhelmed and quit.

This video covers a TL;DR to save you precious study time:

You’ll notice as you start working with AWS that a lot of the other services you actually already know, just with different names (MWAA == Airflow, for example).

Watch the video for a full breakdown of what these core services do, which to safely ignore, and how I suggest learning AWS!

I know I’m a little late to the party on this, but I wanted to share my thoughts on the most recent AI-doomer hype cycle. Specifically:

Funnily enough, I actually heard about this from a completely non-technical friend at dinner last week. She was losing her mind — it was like this was the first time she had ever heard that AI could be dangerous. 🫠

I felt completely unfazed when she explained this story to me. I remember reading Superintelligence way before it was cool, and have been aware that AI could become an existential risk since long before LLMs were called “AI.”

To be clear, I am still concerned about that risk. But what I'm less concerned about is LLMs being the path to get there.

I recognize that I am not an AI researcher, and my take is barely better than that of a layman, but I tend to agree with Cal Newport's perspective that we do not know how to create superintelligent AI, and AI labs using the excuse of recursive self-improvement is a convenient way to sidestep that technical reality.

I’m particularly skeptical when people who stand to benefit from the AI hype cycle participate in AI hype. Telling everyone that your model is terrifyingly powerful can also sound a lot like telling investors that your model is incredibly valuable…

What I AM concerned about is the more mundane ways that LLMs and AI agents can cause problems.

We are not equipped to live in a world with AI agents that can cause security problems like the Hugging Face incident, where deepfakes and fake news reach a level of realism that makes people doubt real evidence, and where organizations hand consequential decisions to systems whose answers sound much more reliable than they actually are.

It's a weird experience for me to be an AI engineer in 2026, but almost by accident. I was just a Machine Learning Engineer who happened to be in the right place at the right time. I’m not an AI bro at all.

In fact, I feel ambivalent about AI (even just the LLM-flavor). I see the immense potential for good, and I like building AI systems, just like I liked building machine learning systems before that (they're really not that different).

But I don't care at all when there's a new model release. I don't buy into the hype cycle in either direction. I'm not scared of existential risks right now, but I also quit my job at Twitch in part because I don’t enjoy coding with AI.

So this is just my gentle nudge to remember that there's room for gray in the world.

I'm not sure if we're doomed. But I am skeptical of anyone who simplifies nuance as a matter of course.

For now, I’d like to see more attention paid to the systems we’re actually deploying: what they’re allowed to do, how we check their work, and who takes responsibility when things go wrong.

Those questions are enough to keep me concerned (and busy) without needing to believe that the next model release is going to end the world.

Forwarded this email? Sign up here.

Follow me on YouTube | Instagram | X | LinkedIn | TikTok

Note: This email may contain affiliate links. If you make a purchase I may make a small commission, at no cost to you. Thank you for your support!