How can I work toward AI alignment as a software engineer?

Jobs in software engineering for AI alignment tend to be competitive, often drawing many applicants because so many people interested in AI alignment are software engineers. Some jobs in this space can be highly valuable toward reducing AI existential risk, but don’t be surprised if you find it hard to get one. Many openings are listed on the 80,000 Hours job board.

To develop relevant skills, try working in industry as a machine learning engineer (where this definitely doesn’t lead to capabilities advances), do courses like fast.ai, or talk to people at conferences to learn more.

To succeed at getting hired, focus on demonstrating exceptional value to companies through skills and projects. It can pay off to focus on a specific company and learn the tools/stack it uses or to build the kind of things that would have been useful to them in the recent past. Applying early lets you reapply later and can teach you a lot about the hiring process.

There are useful projects you can do in your free time. Alignment Ecosystem Development calls can help you find hobby-style projects, or places where a software engineer could usefully volunteer.

Although there’s more funding for AI alignment projects than there once was, earning to give can still be highly valuable, especially if you know of good unfunded people and projects in your personal network.

Training programs for alignment-related software engineering are sometimes run.

For further advice, you could join an AI Safety Quest navigation call, talk to 80,000 Hours, or see the post How to pursue a career in technical AI alignment.