Jane Street Evolves Software Engineering Interviews to Address AI-Assisted Coding

Jane Street is beginning to trial new software engineering interview formats for campus candidates that move beyond traditional coding tasks to evaluate higher-level engineering skills. While the firm maintains its core interview style—where candidates program through problems while talking through their logic—it is adding rounds specifically designed to test a candidate's ability to verify code quality, structure, and intention in an era where generating code has become significantly cheaper.

This image represents the technical engineering content and blog posts produced by Jane Street.

Shifting Focus to Software Design

As LLMs become standard tools for reading, debugging, and reviewing code, Jane Street is adapting its hiring process to reflect these changes. The firm's new onsite rounds for campus candidates lean into higher-level concepts, such as software design and the evaluation of code structure.

The goal is to identify engineers who can make critical decisions about what to build and how to structure it. By focusing on the 'intention' of the code, the firm seeks to find candidates who can ensure that the software produced—whether by a human or an agent—is actually correct and well-architected.

“As generating code has become arguably cheaper than ever, other parts of the job have come to matter even more: taste about what to build, choosing abstractions that last, and making sure what got built is actually correct, among other things.”

— Our software engineering interviews are changing (a little)

Evaluating 'Taste' and Tool Awareness

Jane Street emphasizes that as code generation becomes a commodity, the ability to exercise good 'taste' becomes a primary differentiator. This involves choosing abstractions that last and understanding the specific strengths and weaknesses of the tools being used.

The firm notes that while they find LLMs incredibly valuable, they are looking for engineers with great instincts for building software. They do not require candidates to have a highly worked-out methodology for working with agents, but they do expect candidates to think clearly about when to reach for an AI tool and when to rely on other methods.

Evolutionary and Varied Process

The new interview formats are intentionally kept vague in public descriptions because the firm is still iterating on the specific structures. These rounds are not yet universal across all onsite interviews, and candidates will receive a heads-up on what to expect when they arrive.

For more experienced candidates, the interview process remains more varied. These sessions continue to feature a range of styles, including system design discussions, producing software, and reviewing prior work, ensuring that the evaluation remains tailored to the candidate's experience level.