Understanding AWS IoT Core Device Shadow: Reported, Desired, and Delta States

May 29, 2025 • Engineering  IoT 

I've been doing some work for a client that uses Amazon Web Services (AWS) to manage their IoT devices. Specifically, they use AWS IoT Core Device Shadow to synchronize state between their devices and their cloud applications. In this post, I thought it would be helpful to explain how the device shadow works, focusing on the reported, desired, and delta states—and when devices should report their state or respond to changes. Let's dive in!

Go and Rust for the Small Web: How We Build Faster, Simpler Apps

February 26, 2025 • Engineering 

At Limeleaf, we chose to specialize in Go and Rust for our clients and we've written about why we made that choice. However, now that we're building our own products, we need to develop applications for the web, too.

Why Go and Rust?

September 27, 2024 • Engineering 

I was about 80% finished with this post and I was going down the road of listing out all the great things both Go and Rust offer for us to build and maintain software for our clients and product, but then I threw it all out. I realized that posts like that are a dime a dozen and there are so many resources out there comparing, contrasting, and listing features of both languages, but none of that really explains WHY we, the humans Erik, John, and I, like using Go and Rust. So instead I thought it would be better to personalize this a bit and grab some quotes from each of us about why we like each of them, unedited, unfiltered, and raw!

Safer Binary Decoding in Go

May 30, 2024 • Engineering 

Go is a popular language choice for building web services. Typically, when building those web services, we end up encoding/decoding JSON as the data format. The encoding/json package provides a safe way to turn JSON payloads into Go structs, and vice versa.

However, if we need to handle raw []byte that follow a binary encoding format that is not self-describing, we need to do a bit more work and implement the encoding.BinaryMarshaler and encoding.BinaryUnmarshaler directly. Since we're dealing with []byte, we need to respect slice bounds to avoid triggering a panic() and crashing our service.

Let's look at the two ways we can decode data into Go structs and compare how one way will be safer than the other while yielding the same result. As an added bonus, we'll end up with easier to understand code.