🦸 Community Superhero

Our community superhero this week is Paula Wakabi, founder of Women Innovating with Cloud in Africa. Paula has a gift for creating space where others can shine, from building communities to opening doors to leading women in tech with passion and generosity. Between her speaking engagements and running an AWS user group, she’s always lifting others up and giving back. Thank you, Paula, we should all try to be a little more like you 💙

💯 Spotlight

This week we’re starting off with a piece of content that has been missing for months. Matt Martz published a birthday post (happy birthday, Matt 🎂) showing us in great detail how to build non-trivial serverless MCP servers. Before I continue, this article does not pass the “I wrote this without ChatGPT's help” sniff test, but it provides enough beneficial and novel information I decided to include it AND make it my spotlight for the week. This blog post goes over how to build MCP servers meaningfully using the CDK, gives you a working code repository to build an MCP server that uses Cognito for auth, shows you patterns for triggering AWS services like Step Functions as part of a tool execution, and gives you recommendations for how to group your tools together. It’s thorough, it’s clean, it reads like an AI-assisted article 🙃, but it is very good and worth the read.

🔥 My Favorite Content

“Back to basics” articles are always a joy to read. Typically written by senior+ engineers who have a particularly good understanding of a foundational component of software development, these articles tend to explain hard concepts easily. Seth Orell did a fantastic job of this last week in his article “Queues, Buses, and Streams” where he explained… queues, buses, and streams. It explains what the practical usage is of each one of them, where they fit into your architecture, and even goes into explaining how Kafka fits into all this. Wonderful job on this, Seth!

Years ago, I wrote an article about geo-search using DynamoDB and I admit - it was a pretty naive approach. But it worked for what I needed and people don’t talk about that use case with DDB very often, so I guess it was good enough. Ian Brumby published an article last week that… yup, totally blew my solution out of the water. His post, “Effective handling of geospatial data in DynamoDB,” shows a real mastery of the subject with a production-grade solution. He articulates the problems clearly, shows you the quick solutions, and provides easy-to-understand examples to guide you through.

When I was at the AWS Hero Summit a few weeks ago, I had a conversation with the general manager of Systems Manager (the umbrella of AppConfig, parameter store, and quite a bit more) about storing secrets in SSM vs Secrets Manager. His polite-yet-hesitant responses to me describing my process signaled I don’t do things “the best way.” Which makes Ran Isenberg’s post on the subject incredibly timely. Ran tells us which service you should use in various use cases. He gives clear trade-offs for when Secrets Manager is the right call vs when you should be storing something as a secure string in SSM.

I published a video last week explaining how vector indexes work and how they are used in the world of AI for searching. It’s a really short video, almost 2.5 minutes, but is dense with information using candy to describe how it all works. I’m really proud of how it turned out - the animations took me forever to do, but I felt like it was worth it because it makes the concepts sticky in your brain after you watch it. If you don’t quite know how vectors play into the AI world, this video is worth 2 minutes of your time.

💡 Tip of the Week

We went through a massive MCP hype a few months ago that has since calmed down quite a bit. But it seems to have left an imprint on many developer minds that “I need to make an MCP server for others to consume.” Erik Wilde shared a bit of his industry knowledge from the world of APIs on a pattern he’s seeing repeated with MCP.

🐣 New Releases

Reminder - AWS News is the best source for AWS-related service announcements. For all releases and summaries of what happened, head over there!

Vercel is pioneering the world of FaaS. Last week they began supporting graceful shutdown of functions. This allows you to tie into lifecycle events of your functions to do things like flush logs or close database connections.

The CDK just introduced CDK refactor in preview, which allows you to rename constructs, move resources between stacks, and reorganize your applications without deleting and recreating your existing resources. This is a huge quality of life improvement that will allow your apps to evolve easier.

The AWS toolkit for VS Code now has LocalStack integration. This allows you to more easily test and debug your apps locally. It pairs nicely with the remote debugging feature they announced recently.

Last Words

It was a light AI week, how about that! I’m getting a sense that the flashy hype might be subsiding and now people are turning a little more serious about getting AI agents out to production safely. While we’re still a bit off from that being the norm, It feels like a significant step in the right direction. I’m excited about what the future holds regarding agents and new capabilities for software development.

That’s my take on the week, but what’s yours?

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