Anthropic

What Anthropic does

Anthropic is the AI safety lab behind the Claude model family, founded on January 26, 2021 by Dario and Daniela Amodei, Jared Kaplan, Jack Clark, Chris Olah, Ben Mann, Sam McCandlish and Tom Brown. Its stated mission is to build reliable, interpretable and steerable AI systems, and it publishes heavily on interpretability, mechanistic interpretability, and alignment evaluations. By 2026 it had grown into a frontier lab operating at enormous scale: annualized revenue reportedly climbed from roughly $9B at the end of 2025 to $47B by May 2026, per Sacra. Its own December 2025 research found employees self-reported using Claude in about 28% of their daily work with a roughly +20% productivity boost — the closest thing to a quantified benchmark for how AI-native a lab's engineering workflow really is. For engineers, the work centers on frontier model development plus the safety and "containment" systems that bound Claude agents as their capabilities grow.

What it is like to work at Anthropic

The mission focus on AI safety is genuine and pervasive, and it is the single biggest reason engineers join. But the hours are long: Reddit testimony from staff describes "60-80+ hour weeks were the norm for technical and nontechnical roles alike." Compensation and culture score very high on Blind and Glassdoor, while work-life balance is the weakest area, rated around 3.6/5 per a March 2026 jobsbyculture profile. Glassdoor themes are mixed, including "good salary and benefits, but toxic culture persists" and "great for safety research, but work-life balance suffers." Reviewers and outside commentary also flag a specific hiring profile — one widely-shared post characterized Anthropic as favoring "young, elite-school, Bay Area, math-heavy engineers from top schools." The net picture is a place where believers in the safety mission and long-horizon research thrive, but where standard 40-hour expectations and non-elite backgrounds can feel out of step with the culture.

What Anthropic pays

Anthropic is a top payer, and compensation is consistently rated its strongest dimension on Blind — reviewers put pay and culture at the top even as they put work-life balance at the bottom. The equity story is dramatic: the valuation went from $183 billion post-money at Series F on September 2, 2025 to a $965 billion post-money at Series H on May 28, 2026, roughly a fivefold increase in under nine months, with Altimeter, Dragoneer, Greenoaks and Sequoia leading the later round. Against annualized revenue that reportedly rose from about $9B to $47B over the same window, that valuation implies grants with substantial paper upside for anyone who joins ahead of continued growth. The practical trade-off mirrors the broader frontier-lab pattern: the pay and equity are the compensation for genuinely demanding hours. Candidates should size the grant against a schedule that reviewers describe as 60-to-80-plus hours in busy stretches.

How hiring works at Anthropic

Anthropic's process is long and values-heavy. The Exponent guide documents seven to nine rounds: screening calls, a technical assessment, and two onsite loops. Its take-home assignment was open-sourced in early 2025, giving candidates unusual visibility into what is asked. The igotanoffer Engineering Manager guide reports that questions are explicitly tied to Anthropic's seven core values, probing tensions like "speed versus safety, candor versus diplomacy, and autonomy versus collaboration." That values tie-in is not decorative — it is a genuine filter, and candidates who cannot speak credibly to the safety mission tend not to advance. Because the loop is longer than most FAANG processes and leans on both a take-home and behavioral depth, prep is non-transferable from other companies. Plan for a multi-week cycle and expect to be assessed as much on judgment and mission alignment as on raw coding.

Growth & trajectory

Anthropic's growth is among the fastest in the industry. Headcount roughly doubled from about 2,900 in late 2025 to around 5,000 by mid-2026 — approximately 1.7x in eight months — while valuation went from $183B to $965B over a similar span. Annualized revenue reportedly tripled and then some, from roughly $9B at the end of 2025 to $47B by May 2026 per Sacra. That combination of tripling revenue and near-fivefold valuation growth makes it one of the most richly-capitalized private companies in the world, now valued at or above the public-market AI plays. For an engineer, the upshot is abundant resourcing and clear momentum, but also a young, rapidly-scaling organization whose internal structures are still maturing. The pace of capital is matched by the pace of headcount, which means processes and management layers are being built in real time rather than inherited.

Risks to know

The clearest risks are cultural and organizational. Work-life balance is the lowest-rated dimension, and Glassdoor reviews explicitly call out an elite-school culture that some find exclusionary. The rapid 1.7x headcount growth in eight months means management structures and processes are unproven at the new scale. On the mission side, Anthropic drew safety-community criticism after releasing Claude Gov in mid-2025 and winning a $200 million US Department of Defense contract — a tension for a company built explicitly around AI safety. Research themes the company itself publishes, around model deception and jailbreaking, are a reminder that the underlying technology carries brittleness risks. For candidates, the practical concern is joining a company scaling faster than its internal systems, where the safety mission that attracts people can collide with commercial and defense pressures, and where the demanding hours are a documented rather than incidental feature of the environment.

Who thrives at Anthropic (and who should not)

Thrives: engineers who genuinely buy the AI safety mission, researchers who want long-horizon interpretability and alignment work, and people comfortable with Bay Area working hours and a math-heavy, elite-school-leaning peer group. The environment rewards deep, rigorous, safety-conscious thinking over rapid consumer iteration. Should not join: candidates who need a stable 40-hour workweek, people who want to ship consumer products on tight feedback loops, and engineers uncomfortable with the specific cultural profile that some Glassdoor reviews explicitly flag. The honest read is that Anthropic offers unusually mission-aligned, well-compensated, high-momentum work for the right profile, but it is a poor fit for anyone whose primary goals are predictable hours, fast product cycles, or a culturally broad workplace. Fit here is more about alignment with the safety mission and research temperament than raw ambition alone.

Roles Anthropic is hiring for

The roles Anthropic is most actively hiring right now in our index, with a live count and the salary guide for each:

The full board of open roles — with comp and location on every posting — is at the top of this page.

The signals behind this page

The hiring picture here is read from 240 live Anthropic postings in our index (refreshed weekly); 100% are remote-friendly, and in a recent sample 174 disclose a pay-transparency band. The culture, growth, and interview detail above is researched and cited; the open-roles board is live from our jobs index.

Sources

• Anthropic — How AI is transforming work at Anthropic (Dec 2, 2025)
• Anthropic — raises $65B Series H at $965B post-money (May 28, 2026)
• Anthropic — Series F at $183B post-money (Sep 2, 2025)
• jobsbyculture — Working at Anthropic 2026
• Sacra — Anthropic revenue, valuation & funding
• Exponent — Anthropic software engineer interview guide
• Hacker News — Anthropic take-home open-sourced (Jan 2025)
• Britannica — Anthropic (DoD contract, history)
• Glassdoor — Anthropic reviews