Anthropic manager drops coding interviews as Claude outperforms candidates

Image- Collected

An engineering manager at Anthropic has stopped using coding interviews to recruit for his team, arguing that AI tools such as Claude have made traditional coding tests less effective at evaluating candidates.

Felix Rieseberg announced the decision in a post on X on Tuesday, October 27, saying he had removed coding interviews from his team’s hiring process that month. He clarified that the change applies only to his team, not the entire company, and admitted he has yet to find a suitable replacement.

In a blog post titled “Hunger beats tenure: Advice for junior engineers,” Rieseberg said small coding challenges had previously served as a limited but useful predictor of success. However, he now believes they no longer provide meaningful insight, as Claude can outperform candidates in tasks such as React programming and assembly-language questions.

He added that some of the most effective engineers on his teams perform well regardless of their coding-test results. He has also observed relatively inexperienced developers surpassing colleagues with stronger coding skills and longer careers.

Rieseberg’s advice is aimed at junior engineers who lack professional connections, status or influence. He encourages them to understand their company’s goals, identify an ambitious idea that could advance those objectives, and build a working prototype over a weekend. He cautions against combining too many ideas into a single project.

As an example, he highlighted two relatively new Anthropic engineers, Anthony Morris and Alice Zhao, who developed a prototype graphical interface for Claude Code over a weekend. Its first public version was released about eight days later.

For hiring, Rieseberg currently reviews résumés and portfolios for small prototypes that demonstrate good judgment, empathy and thoughtful decision-making. He has invited others to suggest more effective recruitment methods.

The shift reflects a broader reassessment of technical hiring across the technology industry. Google, owned by Alphabet, is expanding pilot interviews that allow candidates to use its Gemini AI assistant. Google principal engineer Jaana Dogan has also expressed interest in assessing how candidates work with coding agents.

Amazon, meanwhile, has instructed engineers to prioritise customer and business outcomes over simply increasing their use of AI.

However, traditional skills assessments remain common. According to an October 2026 employer survey cited by ZipRecruiter labour economist Nicole Bachaud, 62.5 per cent of employers in the information sector require formal skills tests during initial applications. She said 70 per cent have shifted towards proctored or in-person assessments to reduce AI-assisted cheating.

Bachaud noted that technology companies are changing how they evaluate candidates rather than necessarily abandoning testing altogether. The Decoder also observed that Rieseberg’s emphasis on building projects quickly assumes a level of time and personal capacity that not every junior developer can afford.

Despite these challenges, Rieseberg does not expect traditional coding interviews to return to his hiring approach, writing that he believes the format is unlikely to make a comeback.