Dec 06, 2019

Amazon thinks you can code better

Amazon CodeGuru

Amazon announced CodeGuru, a new tool to automate code reviews. It promises to find your most expensive lines of code and suggest how you can improve them.

How? CodeGuru’s machine learning models are built with data combined from Amazon’s internal codebase and more than 10,000 open source GitHub projects. CodeGuru mines Amazon’s internal codebase for pull requests that include AWS API calls and matches them to relevant documentation to provide you with code suggestions.

CodeGuru looks for resource leaks, potential concurrency race conditions, and wasted CPU cycles. Through its automated reviews, CodeGuru promises to reduce CPU utilization, cut compute costs, and improve application performance.

Why it’s a big deal: We’ve seen a lot of activity in the world of automated code review technology. Most notably, GitHub recently acquired Pull Panda and Dependabot to strengthen its pull request automation features.

While GitHub has the advantage of being home to most of the world’s open source code, Amazon dominates everything related to infrastructure.

Using its infrastructure experitse, CodeGuru differentiates itself from the growing field of automated code review tools by making unique application performance recommendations. As the largest cloud service provider, AWS can leverage its huge developer community and internal best practices to make a clear and measurable impact on developers’ code.

But... notice how Amazon is still dependent on code hosted on GitHub to train its models. Microsoft’s Azure can piggyback off GitHub’s spot at the center of the developer world to drive growth, but AWS will need to be more creative in capturing developer mindshare.

CodeGuru only supports Java applications today. That will likely expand rapidly as Amazon looks to defend its top spot in cloud-based development.

Want to get more of these in your inbox?

Subscribe for weekly updates from the Software team.