🦸 Serverless Superhero

Our serverless superhero this week is Arjita Mitra, DevOps engineer at PrimEra Medical Technologies, AWS Community Builder, and quite a remarkable person! She is a recipient of the All Builders Welcome grant, an AWS New Voices speaker, and shares some fantastic content online. She has wonderful optimism and energy, and brings a positive vibe along with her to the community. Thank you for everything you do, Arjita!

💯 Spotlight

I love API design. Whenever I see posts that improve DX around APIs or offer some clever way to do something, I’m all over it. Anne Stein published a blog post that showed us how to easily make mock APIs with the CDK using nock. It’s an incredibly clever solution that offers so much pragmatic value. Her post shows you how to capture traffic and turn it into a mock server with basically no effort at all. Love it!!

🔥 My Favorite Content

AWS Amplify is a great service for hosting websites. I’ve used it for Ready, Set, Cloud since its inception and have used it for at least a dozen other projects. But what I’ve found is that there’s certain things that just make the build process break over and over again. Fortunately for us, Felipe Malaquias published a blog last week showing how to use Docker in AWS Amplify builds. I’ve struggled with this a few times and usually find some hacky workaround. But Felipe shows us the real answer in his post.

Remember the crazy amount of hype a couple of years ago when Bun (a JavaScript runtime) was released? Everybody claimed it was was so much easier and faster than Node, plus it had native support for Typescript. But Lambda never supported it as a runtime despite the performance claims. Well, Jason Butz decided to take that into his own hands last week and show us how to use Bun as a custom runtime for Lambda. Not only does he walk us through how to build and deploy the runtime, but he also shows us what changes with your function outputs and compares the performance with the Node runtime - with interesting results.

In my opinion, AWS Step Functions doesn’t get the love it deserves - especially with recent releases like JSONata support and variables. That said, Damien Jones had a wonderful writeup last week showing us how to use Step Functions and JSONata for low-code S3 key validation. I like this post a lot because it shows us practical JSONata expressions and how he used them in a real project. This is a massive accelerator for people trying to figure out how to get started with it. Great work Damien!

This next post is not serverless related, but it uses something I think many of take advantage of in our automated tests. Jonathan Geiger published a blog on how to block requests with puppeteer. Puppeteer is a JavaScript library that runs headless browsers to automate tasks. In many instances, you don’t need the webpages you’re testing to download static assets or make requests to user-tracking endpoints. This is exactly what Jonathan shows us how to avoid to make our tests run faster and more efficient than ever.

My last pick of the week I have mixed opinions on. Jayant Harilela published a blog post titled “How I built an AI business validator to generate 300 blog posts for me every day” and while I disagree wholeheartedly with that premise - the build is actually pretty cool. He talks about the tool he built that gives you a yes or no answer to your business idea along with some info about how it could possibly do and what the addressable market is. This is a great way to use AI and he’s doing a significant amount of work behind the scenes. I’m impressed by the project but want to emphasize my (strong) opinion that you should not use AI to create and publish blog posts. Keep content human.

💡 Tip of the Week

As new things continue to emerge in the AI frontier, it’s becoming easier to think about the possibilities in the not-too-distant future. I had a great conversation with Sterling Chin last week that he summarized in a LinkedIn post.

🐣 New Releases

Amazon announced SaaS Manager for Amazon CloudFront, which is a new CloudFront feature aimed at providing a unified experience across websites for SaaS providers. Seems like it has some promise!

Meta Llama 4 models are now available in Bedrock. The service is bringing in Llama 4 Maverick 17B and Llama 4 Scout 17B which are both multimodal models with over 1M token context windows.

The Amazon Q Developer CLI now supports MCP servers. This greatly enhances the capabilities of that CLI.

A new Amazon Nova model was released last week. This model, Premier, is their most capable model, and is capable of multistep planning and precise execution across tools and data sources.

In a big announcement, CloudWatch launched tiered pricing and additional destinations for Lambda logs. Based on volume, your log pricing can fall as low as $.05 per GB AND you can ship logs off to S3 or Amazon Data Firehose. Wow!

Last Words

Please join me in a big welcome to the new voices of this newsletter! I love seeing new opinions and styles in here. It’s a great way to learn and adopt different ways of thinking. Please give all of our contributors a follow!

If you’d like to make a recommendation for the serverless superhero or for an article you found especially useful, send me a message on Twitter, LinkedIn, or email.

Happy coding!

Allen