We've been using AWS Lambda function to asynchronously render PDFs, transform data or execute on different pipeline triggers with success for a few years now. The horizontal scaling that avoids having a pressured queue during high-traffic spikes is pretty nice.

Another case we've had success with is transforming and relaying webhooks from different services (relaying events into Datadog for services that don't have a native integrations) or acting on occasional webhooks that don't happen too often (creating a default set of labels on a new Github repo).

If you're doing anything asynchronously, it's pretty nice. Be aware that the amount of memory you allocate to your functions is directly proportional to the CPU amount (and price) you're given. There's also the caveat that invocations perform _at least once_ but not necessarily once. A few months ago we saw a significant increase in repeat lamdaba invocations/retries even when the original execution was successful.