Back in July, Microsoft revealed how it is heavily incorporating AI into its development workflows. It talked about its in-house AI-powered coding assistant being leveraged for code reviews of over 600,000 pull requests (PRs) per month, encapsulating 90% of all PRs generated at the company. Microsoft is also integrating AI technology deeply into its consumer-facing development tools. Now, a Microsoft veteran has revealed even more ambitious plans in this area.
In a detailed LinkedIn post, Microsoft Distinguished Engineer Galen Hunt has proclaimed that his goal is to replace every single line of C++ and C code at Microsoft with a combination of Rust and AI assistance by 2030. Hunt"s mission is to enable a single engineer to write a million lines of code per month, which is a pretty arbitrary (useless?) metric to evaluate productivity, as recently noted by Linux" creator Linus Torvalds in a different context.
Hunt has further highlighted a "powerful code processing infrastructure" that enables the firm to deploy AI agents guided by defined algorithms to make changes to existing code at scale. To make this tooling even better and translate C/C++ systems to Rust, the veteran"s team is hiring a Principal Software Engineer with at least three years of experience in writing system-level Rust.
The Rust programming language has been gaining popularity at Microsoft over the past couple of years. The tech giant has been encouraging driver development written in Rust for better security. Meanwhile, Hunt himself has worked for almost 30 years at Microsoft and is currently a member of the Future of Scalable Software Engineering group in the EngHorizons organization in Microsoft CoreAI. His team"s responsibility is to develop new tools and technologies that eliminate technical debt at the company, and eventually, for the rest of the industry too.
The response to Galen Hunt"s post on LinkedIn has been mixed. People are mostly questioning the move to Rust, with the engineer defending this choice by highlighting the language"s built-in memory and concurrency safety mechanisms.