Did the HF PM loop about six weeks ago. Got an offer, accepted, haven't started yet. Writing this while it's fresh because the HF PM interview is different from any other PM loop I've done and the prep advice out there is basically useless.
Context: I'm a senior PM, 8 years in, mostly B2B SaaS before this. HF was a left turn for me, intentionally.
Here's what I found:
They don't do classic PM frameworks. No "design an alarm clock for a blind user" stuff. No CIRCLES method. None of it. If you prep that way you'll sound like a PM interview prep course, not a person.
What they actually asked: "What's broken about how developers discover and evaluate open-source models today?" This is a product problem they live with. They wanted my actual take, not a consultant's diagnosis. "Pick a product decision HF made that you think was the wrong call. Defend a different approach." This one is a test of whether you've actually used the product and have opinions. "How do you prioritize between building tools for the research community vs. for enterprises trying to deploy models?" This is live tension for them. Wrong answer: try to serve everyone. Right answer: show that you understand the tradeoffs and have a point of view. Behavioral: one question about working in a fast-changing environment, one about a product decision you made that you later regretted and what you'd do differently.
The metrics conversation: They asked about a product I'd worked on and how I measured success. Push back if the obvious metric is bad. They wanted to see that I'd challenge a vanity metric. This matters at a company where the community-driven product has hard-to-measure outcomes.
What to prep: Actually use Hugging Face. The Hub, Spaces, model cards, the inference API. Have opinions. That's the whole game.