🦸 Community Superhero
Our community superhero this week is Piyush Sachdeva, engineer, developer educator, and SaaS builder. Piyush runs a YouTube channel called Tech Tutorials with Piyush with hundreds of videos on DevOps, cloud computing, and trends in our industry. He’s also built multiple SaaS products that help content creators grow their skills and following. Thank you so much for everything you do, Piyush!
💯 Spotlight
There’s no better feeling as an engineer than ripping out hundreds of lines of code. Simpler the better, I always say, and that’s exactly what happened with MCP back in July. Maria Paktiti caught us up last week on the most recent version of the spec, which rips out WAY more than it added. If you haven’t been keeping up, this is totally worth the read, especially for this audience. MCP is stateless now, and they’ve made significant leaps toward making it feel standard and familiar. Huge update!
🔥 My Favorite Content
We talk every week about how software engineering has fundamentally changed. Like, every part of it. It can be hard to keep up with what’s available and how to tune our working styles to maximize results from the available tooling. Baris Taze published his opinionated design and architecture guidelines last week, which are a tremendous help. I have a lot of respect for Baris, and seeing his habits in written format with easy explanations of why he does things is just pure gold. He has the full details in GitHub, which you should definitely read yourself, but you could also hand it to your agent too to start building like he does.
Baris talked about honing the coding side, which is obviously important. But perhaps more important is how we use AI in the entire development process, because the biggest long-term gains are there. Nic Alpi walked us through what he does for his team, informing us that if you’re measuring AI by speed, you’re measuring the wrong thing. His article is a great story about how he shoots for 1.25 or 1.5x developers, not 10x. What he wants isn’t necessarily speed, it’s solving problems better for customers. I really appreciate the take in this article, as someone who has lived the scoping and planning life. Great read.
We’ve talked about coding with AI and development processes with AI. But what about debugging with AI? Arguably the thing you spend the most time on as an engineer, how do we do that better? Afam Onyimadu took us through a short case study of Linus Torvalds debugging a Linux kernel issue with AI and made some astute observations. He points out that the Linus story isn’t new, but people have been generalizing what he did. Afam calls out the main takeaway people keep missing when they read the story, which is mainly “let AI search, you make the decisions.” As we all know, AI can be confidently wrong, and that backs us into a corner all too often.
Enough about AI, let’s talk about testing. Something we all need a ton of because AI is doing everything for us. Daniel Balcarek wrote about designing realistic performance tests last week in impressive depth. He goes over the types of performance tests, how to design them for a realistic expected load, what metrics to track, and a lot more. He has friendly illustrations throughout the article that help explain the content really well. While yes, you could give this to your agent so it writes better tests, it’s worth the time to read it yourself.
💡 Tip of the Week
We’re seeing more and more LLM “accidents” come to the public eye these days, and they’re all over the board. But no matter what type of accident it is, we seem to be pointing a finger of blame at the wrong thing. Ross Turk made a great observation last week that I had to share.
Last Words
My goodness, what a week! If you were feeling confident in your AI skills, how do you feel after that? I have a lot of thinking and rethinking to do after this past week’s content, and I’m here for it. Hopefully y’all feel the same way.
What did I miss? What made you nod along (or 🙄)? Hit reply if you’re reading the email. Prefer socials? Ping me on Twitter, LinkedIn, or email.
Happy coding!
Allen