Jump to a section

The theme to many of the newsletters I’ve been getting recently is unrentlingly about AI but also some backlash and criticism about how AI is implemented. You’ll see in the AI and General Tech sections a lot of discourse about not only how we should be questioning model context protocol (MCP) but also some differing opinions about how AI is affecting the jobs and thought processes of engineers.

This isn’t the end or the beginning of the discourse but seems interesting that these thought pieces are still coming out with interesting flavors and takes, most more nuanced than my own take on AI’s pressure on engineers even as we iterate our thoughts as a collective. But there’s also a big part of me that wishes that wasn’t all we talked about. Hopefully below you’ll find some diversity in the discourse and some tips/tricks/stunts for your engineering.

🤖 Android

  • In recent times, I was part of a project that looked at using ProtoBuf for service layers that communicate with frontend clients. This is a good breakdown that pits JSON against ProtoBuf with use cases. ProAndroidDev
  • I feel like this is something that I should have know about before this article but here’s a brief dicussion and implementation walkthrough for Room, local/session storage for Android (I think? Tell me if I’m wrong, please). Nathan Kayumba
  • In case you want to know what Google thinks your Android apps should look like, Material Design 3 leaked early. Ar Technica
  • Coroutines. They can be a mud bog so Philipp Lackner tells you how to make yours better. Philipp Lackner
  • Not one but TWO videos about how to use JetBrains AI Assistant to help you code. Code with the Italians , Kotlin by JetBrains

🍎 iOS

  • Get Picture-in-Picture going for your iOS app without feeling like you have to hack the system. Artem Novichkov
  • A multi-part series in understanding targets, schemes, and build configurations in Xcode, an admitted blind spot for me (just work Xcode, geez). Artem Novichkov
  • Regex is the devil’s work and you can’t tell me otherwise. But here’s everything you need to know about regex in Swift. Guy Lapalme
  • Apple and Anthropic are teaming up to bring Claude into Xcode. Maybe better than the deal they made to build Apple Intelligence with OpenAI? TechCrunch

🕸️ Web

  • Next.js -> Ruby on Rails + Inertia.js (what is old is new again). Hardcover Blog
  • There is an eternal struggle for whether to build apps in native iOS/Android versus using a hybrid or webview approach. Webviews are cheaper (iOS and Android deployment/labor costs are higher) but never quite feel like an app build exclusively in Swift or Kotlin. Until now? Shopify feels like they cracked the code on makeing webviews feel like native apps. Shopify Engineering Blog
  • Make your web app faster and more performant in 2025 with these 14 tips! Sia Karamalegos
  • The WCAG 3.0 proposal with a scoring model kind of gamifies a11y and I’m here for it. Smashing Magazine
  • Yet Another React Alternative as people quietly want to move away from React. Nue
  • All the features that made their way to browsers in April 2025 (at least the ones that Google wants you to know about). web.dev
  • A trip down memory lane for the olds: image maps. CSS Tricks

☕️ Java/Kotlin Backends

  • Domain-driven design seems to be all the rage as the pendulum swings back away from “the client is so smart and powerful” to “the client should be dumb af so let the backend tell it what to do.” JetBrains digs into a Kotlin solution using Ktor. JetBrains Blog: Kotlin

🏗️ DevOps/Infrastructure

  • Just because one thing caused a SEV-1 and another thing caused a SEV-3 doesn’t mean the work for the SEV-3 isn’t important. Surfing Complexity
  • I’m torn on this article about frontend observability being reheated real user monitoring. I understand where they’re coming from but, also, having insight into how a client behaves independent of the data that comes from the backend seems important and is sometimes mysterious. RedMonk
  • Sometimes I need a remdial class to remind me of stuff I should know like 5 Core Distributed Concepts Every Developer Should Know. SWE Quiz
  • A Docker MCP Catalog and Toolkit could be super interesting. Docker

🦾 What We’ve Decided to Call AI

  • I like this for two reasons. (1) Acting like an assistant is all I want from AI and I like the idea of doing a virtual walk and talk where I can ask, “What’s next?” and (2) it’s named like it’ll start inserting tasks like “Get this banana in the bottle ” into my day. eyaltoledano (Github)
  • Anthropic looks at how engineers are using Claude. Do you see yourself? Anthropic Research
  • Cloudflare shows off how you can use them for implementing your brilliant MCP server. The Cloudflare Blog
  • A couple of people have some criticism for MCP and possible alternatives. Let the backlash begin? Tim Kellogg , Raz Holm
  • A visual glossary of AI infrastructure terms by your friends at ByteByteGo. ByteByteGo
  • AI means no one codes here anymore, a positive spin on how AI is going to help the engineer and engineering practices. Alex Ghiculescu
  • I love a deep dive and this one starts to clear up what has been typically a black box for me: NodeJS. DeepIntoDev
  • Vibe Coding a 3D multiplayer game in 15 minutes. I mean, it’s not exceptional looking but it works? How I AI

🌪️ Engineering and Team Management

  • Use reference-class forecasting to better launch big projects. Use what you did before to help predict how you’ll do it this time. Requires that you know how you did it before and/or have data about that. Smashing Magazine
  • A good reminder that, when working at large tech companies, the goal is not to make a perfect little thing but to just get the thing done. Sean Goedecke

🧑🏽‍💻 Technology and Other Stuff

  • As a person that just recently spun up a blog using Hugo, a minimum viable blog is very appealing to me. Carl Öst Wilkens´ Blog
  • It’s not just you. Jobs are feeling worse. Capitalism, amirite? Pluralistic
  • Jobs overall are growing but tech jobs are shrinking. I wonder. The Register
  • If legacy code is generally characterized by code that was created by someone that doesn’t work here and that no one is still around to understand, does that mean code generated by AI (who doesn’t have the capacity to actually remember past experience and context) is immediately legacy code? Text Incubation
  • Learn more languages. Just generally a good rule of thumb for your brain. But also knowing more programming languages is better for your career and can shift your understanding of your main language. Architecture Weekly
  • In a time where I may be thinking about what kind of comp I want, I was intrigued by this “ultimate” guide to negotiating your comp. Lenny's Newsletter
  • The Software Engineer Who Knew Too Much: the paranormal mystery about how your vocation makes you see terrible things that no one else sees. Not a Shelf
  • Tech companies that bleed talent to replace them with AI risk their market share being taken away as people with the same tools can build things with their earned domain knowledge. Tim J williams
  • Things I like about this “day in the life of a productive engineer” include: time spent easing into the day, taking time during the day to decompress, looking out into the river, the desk setup seems to suit him. Things I did not like: running in Converse. Don’t do this. Respect yourself and your feet. Daniel Atitienei