Workday CTO goes back to engineering

Peter Bailis spent less than a year as chief technology officer at Workday, then left in March 2026 to become a member of technical staff at Anthropic, a job centered on writing reinforcement learning systems instead of running a corporate technology organization. (digitrendz.blog) That is a move from the executive floor to the engine room. At Workday, Bailis’s official bio said he was responsible for technology strategy and architecture across the company’s product suite, including Workday Illuminate and its artificial intelligence agents. (workday.com) By April 2026, Workday’s main leadership page no longer listed Bailis and instead showed Gerrit Kazmaier as president of product and technology alongside chief executive Carl Eschenbach. That is the public sign that the org chart changed after Bailis’s departure. (workday.com, workday.com) Anthropic is not a normal software company hiring one more engineer for a mature product line. Its company page says it is building reliable, interpretable, and steerable artificial intelligence systems, and its research page is still publishing fresh work on model behavior and agent safety in April 2026. (anthropic.com, anthropic.com) Reinforcement learning is the part of artificial intelligence where a model is trained with rewards and penalties, like teaching a dog with treats instead of handing it a rulebook. DigitrendZ reported that Bailis’s new Anthropic role is focused specifically on reinforcement learning engineering. (digitrendz.blog, thenextweb.com) That makes the title drop easier to understand. If the hardest bottleneck in frontier artificial intelligence is improving the training loop itself, then a senior builder can have more impact tuning that loop than supervising a few thousand people from a planning meeting. (anthropic.com, anthropic.com) Bailis’s background fits that kind of hands-on work better than a standard enterprise executive résumé. Before Workday, he led artificial intelligence for data at Google Cloud, founded Sisu Data, and taught computer science at Stanford while co-leading the Stanford DAWN systems research project. (workday.com, stanford.edu, stanford.edu) Anthropic has also been building out the opposite side of the ladder at the same time. In October 2025, it hired former Stripe chief technology officer Rahul Patil as Anthropic’s own chief technology officer, which means the company is adding executive management even as it attracts people who want to go deeper into individual technical work. (anthropic.com) The result is a small but telling picture of the artificial intelligence labor market in 2026. In a field moving this fast, “member of technical staff” at a frontier lab can now outrank “chief technology officer” in the one currency engineers care about most: proximity to the code, the models, and the next breakthrough. (anthropic.com, digitrendz.blog)