The thesis: the coding test is no longer useful
Felix Rieseberg, the Anthropic engineering manager who led the development of Claude.ai and Claude Code Desktop, has decided to abandon the classic coding interview. His argument, as reported by The Decoder, is that small coding exercises no longer make sense: Claude handles them better than candidates do. For Rieseberg, assigning a small React challenge today is equivalent to quizzing a candidate on Assembly. In his teams, he adds, a good score on the tests no longer predicts who will become an excellent developer.
Who counts as a 'junior' and why it pays off
The manager redefines the 'junior' figure: it does not necessarily denote someone who is just starting out, but any engineer lacking status, power or an internal network at the company. According to Rieseberg, newcomers have a structural advantage because they notice problems that veterans have accepted as given. He cites the organization theorist James March, who wrote in 1991 that organizations depend on a steady flow of naive people, and recalls Patrick Collison's invitation to new hires to flag what seems strange.
The personal case dates back to his time at Notion: he was asked to design an offline architecture that was internally considered impossible, because the data model was too complex and CEO Ivan Zhao's rejection was taken for granted. The technical obstacle was real; the founder's resistance was not: no one had asked him. Rieseberg argues that the problem for newcomers is that no one listens to them, because organizations were built around a few decision-makers when execution was expensive.
The weekend prototype as the new résumé
With AI making execution cheap, Rieseberg's advice is to understand the company's goals and attach yourself to the parts that generate value: a useful starting point is to know your boss's goals. Then choose an ambitious idea, build a working prototype over a weekend, and skip the cycle of specifications, approvals and lobbying. Better one good idea than three good ones hidden among twenty bad ones; a prototype that mixes a programming language, backend and mobile apps is bound to fail.
Rieseberg distinguishes between two types of engineers: those who work on user-facing products bring natural empathy for those outside, while those who work on infrastructure or internal tools benefit from not knowing why something should be impossible. He cites two engineers who were then relatively new at Anthropic, Anthony Morris and Alice Zhao, who built the graphical interface for Claude Code over a weekend: the first public version arrived about eight days later. Today he looks in résumés for small prototypes that demonstrate taste, empathy and principles-based thinking.
What remains to be verified
The picture has two sides. On the one hand, AI appears as an equalizer that allows newcomers to bypass gatekeepers who have gone blind to their own company's problems. On the other, it shifts the bar for entry-level roles from diligence and craft toward entrepreneurship, product sense and taste—skills that universities teach little of and that companies traditionally attributed to experienced product managers or founders. The formula 'hunger beats seniority' is motivating, but it presupposes a hustle mindset that not all new developers can bring. It remains to be seen how much this approach is already widespread practice at Anthropic or remains the position of a single manager, and how 'taste' is concretely measured in a résumé.
