Anthropic

Anthropic is the most sought-after employer in tech right now. A near-trillion dollar valuation, Andrej Karpathy just joined the research team, engineers leaving OpenAI and DeepMind at a rate of 8-to-1 — the pull is real. Compensation packages for senior roles frequently exceed $250K, with some pushing past $850K all-in.

Getting in is hard. Not just because of the technical bar — but because there's a layer of the process that most candidates have never encountered anywhere else, and it trips up a lot of people who are otherwise very qualified.

This is what the process actually looks like.

Who they're hiring

Anthropic has around 3,000 employees. About 1,000 joined in the last six months. They're not just hiring researchers and engineers — they're building out sales, legal, finance, HR, and marketing teams aggressively. If you have a serious background and the mission resonates, there's probably a role.

The companies they're competing with for talent: OpenAI, Google DeepMind, Meta AI, and to a lesser extent top-tier product companies like Stripe or Linear. People who get offers from all of them are increasingly choosing Anthropic — 80% two-year retention, highest in AI.

The process

Expect up to five rounds, depending on the role:

1. Recruiter screen — standard, 20-30 minutes
2. Hiring manager conversation — first real signal on whether you're a fit
3. Skills/technical assessment — role-dependent, more on this below
4. Culture interview — the distinctive one
5. Final rounds — additional interviews for senior or leadership roles

Two things that are non-standard:

You sign an NDA before speaking to anyone outside the recruiting team. This isn't a red flag — it's Anthropic's standard operating procedure, likely related to how much internal information interviewers share about research direction and products. Sign it and move on.

No AI in interviews. They prohibit AI use in assessments unless explicitly told otherwise. This is worth taking seriously — the culture interview in particular is designed to surface genuine thinking, and they're good at telling when answers are rehearsed or synthetic.

The culture interview

This is what makes Anthropic different.

Most companies have a culture fit conversation that's basically a formality — 30 minutes, a few softball questions about teamwork, boxes get checked. Anthropic's culture interview is not that.

Candidates describe it as feeling more like therapy than a job interview. It can come from someone in any department — a research scientist, a finance person, an ops lead. If they give you a low rating, you're typically out, regardless of how the technical rounds went.

What they're actually testing for:

Intellectual independence. They want skeptics. If you come in saying "I love everything about Anthropic's approach," that's a yellow flag, not a green one. If you've thought critically about how the company is pursuing safety — and have a view that differs from the mainstream — say so. They're hiring people who push back, not people who flatter.

Values under pressure. The classic line of questioning: describe a professional ethical dilemma you faced. How did you handle it in the moment? How do you feel about it now? They're not looking for a clean story where you did the obviously right thing. They're looking for genuine discomfort — the evidence that the right choice cost you something, and you made it anyway.

Unusual beliefs. Daniela Amodei described this directly in a podcast last year: "We ask people, what are some unusual beliefs that you hold, and how have you defended those beliefs in uncomfortable situations?" They're not looking for a particular worldview. They're looking for the capacity to hold one — and defend it when it's unpopular.

Intellectual honesty about AI risk. You don't need to agree with every position Anthropic takes. But if you think AI safety concerns are overblown or aren't genuinely motivated by this stuff, that's going to show, and it's probably not the right place. If you have nuanced views — including disagreements with how Anthropic operates — that's actually good.

What trips people up:

Rehearsed answers. The interviewer will cut you off when the answer stops yielding new information. They're explicitly trying to get signal, not hear a performance.

Being too focused on near-term or "safe" AI risks. One researcher who applied described focusing on chatbot relationship dependency as a risk he cared about — the interviewer seemed uninterested. The questions are operating at a different level of abstraction.

Trying to figure out what they want to hear and saying that. They're specifically trained to detect this. The prep coaches who work with Anthropic candidates all say the same thing: be genuinely yourself, with genuine views.

How to actually prepare:

Think through three things before you walk in:

1. A real professional ethical dilemma — not a humblebrag, not a manufactured one. Something that cost you something to navigate correctly. Be able to talk about how you feel about it now, not just what you did.

1. Two or three beliefs you hold that most people in tech don't. Write them down. Be able to say them out loud without hedging.

1. Your honest view on Anthropic's approach to AI safety — including what you think they might be getting wrong, or where you're genuinely uncertain. If your honest answer is "I agree with everything," you probably haven't thought about it hard enough yet.

The technical bar

Varies a lot by function.

Research roles (alignment, interpretability, capabilities, policy) require serious depth. Expect real problems, not "tell me about your background." The bar is high and the assessment is hands-on. If you're applying for a research position without a strong ML foundation or relevant published work, it's a long shot.

Engineering roles (infra, product, full-stack) are rigorous but not unusually different from top-tier tech. Strong fundamentals, systems thinking, ability to work at scale. The culture interview matters here too.

Commercial and operational roles (sales, partnerships, legal, finance, HR) have a lower technical bar but the culture interview matters just as much, sometimes more. They're hiring people who can represent the mission credibly in conversations with enterprises and governments. That's a real filter.

Compensation

Senior roles: $250K-$850K+ depending on level and function. The equity component is significant given current valuation, but carries the usual startup concentration risk. Early employees could see life-changing outcomes at IPO — the company is reportedly expecting $50B+ ARR by end of this year.

Peter Bailis left his CTO role at Workday to become a "member of the technical staff" at Anthropic. Title cuts for meaningful equity and mission are common.

Worth applying?

If Anthropic is on your shortlist, the process is harder than most but the culture interview is navigable if you approach it honestly. The people who get tripped up are usually the ones who try to game it rather than just... actually having genuine views about the things they're asking about.

The company is hiring a lot of people right now — not just researchers and engineers but across every function. If you're tracking their open roles, the volume of postings has picked up significantly in the last few months as they scale into a much larger commercial operation.