Anthropic · Primly Community

Anthropic interview rejection post-mortem, what I'd change if I could go back

frontend_fran · 5 replies

Got the rejection email two weeks ago after completing the full onsite for a senior frontend/product engineer role. Took a week to stop being annoyed enough to write this clearly. Here's what I think actually happened.

Background: 4 YOE, frontend/React, some TypeScript infra experience. Applied through LinkedIn, got a recruiter ping within two weeks. The process was: recruiter screen, one technical phone screen, then a 4-round onsite (two coding, one system design, one cross-functional).

What I think I did wrong:

The system design round. This is where I lost it. The prompt was around designing a real-time collaborative UI feature. I know this space reasonably well, but I went too narrow too fast. I dived into component architecture and state management before I'd adequately covered the backend contracts and data flow. The interviewer kept asking me to zoom out, and I did, but the recovery was visible. Lesson: for Anthropic specifically, even for product engineer roles, they seem to want you to be comfortable at the full-stack system level before you get to the UI layer.

The behavioral component. I gave decent answers but I think I played it too safe. I picked stories that showed solid execution rather than stories that showed independent judgment in ambiguous or high-stakes situations. For a company that talks a lot about thoughtful decision-making under uncertainty, I probably should have leaned harder into moments where I had to navigate actual uncertainty, not just deliver cleanly.

What I got right: The two coding rounds felt okay. Both were practical, not leetcode-hard. One was more of a debugging/refactoring exercise on a realistic codebase, the other was building a small feature end-to-end with good test coverage. I think I did fine there.

Feedback I actually got: Almost nothing useful from the recruiter. Standard 'we decided to move forward with other candidates' language. One line about 'technical depth at the systems level.' So that tracks with my read on the system design round.

Would I reapply? Probably in 12 months if I spend that time doing more backend-adjacent work. I came in thinking 'senior frontend' meant they wanted a strong frontend person. I think they actually wanted a senior engineer who happens to work on the frontend, which is a different ask.

The process was well-run and the interviewers were thoughtful. The rejection stings but I don't have complaints about how they handled it. Just posting this because I would have wanted to read something like this before I went in.

5 replies

recruiter_rita

The gap you identified, 'strong frontend person' vs 'senior engineer who works on frontend', is something candidates consistently underestimate at AI labs. These companies tend to have smaller eng headcounts than their product surface would suggest, so each IC needs to be more full-stack capable. When they say product engineer they often mean closer to the second definition. Worth asking the recruiter explicitly which way they define it before you hit the onsite.

growth_gabe

The behavioral observation is sharp. I've noticed that companies built around 'doing something important' tend to want stories about times you pushed back, changed direction, or made a call others disagreed with. Safe execution stories don't differentiate you because everyone in the pool has those. What did YOUR judgment look like under real pressure.

frontend_fran

Yeah I had those stories, I just didn't reach for them. I defaulted to 'here's a successful project' mode instead of 'here's a moment where I had to figure something out without a clear answer.' Different framing, same underlying experience, but the second probably lands better in that context.

sdr_sky

Appreciate you posting this. I've been lurking the Anthropic threads trying to figure out if they're worth targeting. Coming from a non-eng background so this process is very foreign to me but the behavioral insight applies everywhere. Saving this.

nonprofit_nia

The 'zoom out before you zoom in' lesson on system design is one I keep seeing in post-mortems from smart candidates who failed design rounds. I wonder if it's something that's almost impossible to learn without actually failing it once. You don't know you went too deep too fast until you see the interviewer's face.