According to Beating, at a Peking University discussion, Moonshot AI president Zhang Yutong disclosed operational details of the 300-person company. The firm has no job titles and maintains an extremely flat structure, with no more than one or two organizational levels between any two employees. Job boundaries are intentionally blurred: employees can transition between pretraining and post-training, algorithms and data, or marketing and model evaluation.
For hiring, Moonshot AI uses an AI interview system that ignores educational background and focuses on two traits: ability to propose original ideas and willingness to iterate obsessively on concepts. The system tracks iteration count, different approaches tested, and timing of attempts to identify candidates with abstract thinking and persistence.
Related News
Andrew Ng: “AI won’t trigger a mass wave of job losses,” while software engineering hiring remains strong
Walmart Cuts 1,000 Corporate Jobs in Tech and AI Consolidation
Anthropic talks $30 billion funding round, valuation could break $900B—challenging AI market-cap peak