Anthropic manager removes coding tests from his team’s hiring
Felix Rieseberg says he now looks for prototypes in candidates’ portfolios, though he has not settled on how to assess engineers who build with AI.
Anthropic engineering manager Felix Rieseberg says his team no longer uses coding interviews because Claude can outperform candidates on the tasks involved. He described the decision in a blog post dated 27 October, arguing that small programming tests have become less useful for identifying effective engineers.
Rieseberg wrote that coding challenges had previously offered a limited indication of how candidates would perform on his teams. Now, he said, the people making the biggest contributions succeed regardless of their test scores. He also argued that familiarity with frameworks and libraries matters less than before, allowing some newer engineers to surpass more experienced colleagues.
He has not yet established an interview method for assessing how well candidates build with AI. Instead, he examines résumés and portfolios for small prototypes that demonstrate judgment, empathy and clear principles.
As an example, Rieseberg cited Anthony Morris and Alice Zhao, two relatively new Anthropic engineers who created a graphical-interface prototype for Claude Code during a weekend. According to his account, a public version followed roughly eight days later. His biography lists engineering leadership work on Claude.ai, Claude Code Desktop and Anthropic’s Cowork tool.
