| Role family | Postings | Median midpoint |
|---|---|---|
| engineering | 172 | $402,500 |
| support / ops | 75 | $265,000 |
| finance / legal | 74 | $285,000 |
| sales | 71 | $335,000 |
| data / ml | 49 | $595,987 |
| product | 43 | $345,000 |
| other | 39 | $294,645 |
| security | 32 | $362,500 |
| marketing | 25 | $320,000 |
| design | 9 | $345,000 |
| Title | Location | Low | High |
|---|---|---|---|
| Research Engineer, RL Scaling Science | London, UK | $502,236 | $857,149 |
| Engineering Manager, GPU (ML Accelerator) | San Francisco, CA | New York City, NY | Seat | $500,000 | $850,000 |
| Pre-training Data Infrastructure Engineer | San Francisco, CA | $500,000 | $850,000 |
| Research Engineer, Chip Design RL | San Francisco, CA | $500,000 | $850,000 |
| Research Engineer, Code RL (Reinforcement Learning) | San Francisco, CA | New York City, NY | $500,000 | $850,000 |
| Research Engineer, Computer Use | San Francisco, CA | New York City, NY | Seat | $500,000 | $850,000 |
| Research Engineer, Cybersecurity RL (Reinforcement Learning) | San Francisco, CA | New York City, NY | $500,000 | $850,000 |
| Research Engineer, Machine Learning (Reinforcement Learning) | San Francisco, CA | New York City, NY | $500,000 | $850,000 |
| Research Engineer, Machine Learning (RL Velocity) | Remote-Friendly (Travel-Required) | San Fran | $500,000 | $850,000 |
| Research Engineer / Performance Engineer, RL Distributed Systems | San Francisco, CA | New York City, NY | Seat | $500,000 | $850,000 |
| Research Engineer, Post-Training Model Evaluations | San Francisco, CA | Seattle, WA | $500,000 | $850,000 |
| Research Engineer / Research Scientist, RL Frontiers | San Francisco, CA | New York City, NY | Seat | $500,000 | $850,000 |
Across 589 postings that state a pay range, the median midpoint is $347,500. The middle 50% sits between $282,500 and $402,500, and the 90th percentile is $458,000. All figures are annualised and converted to USD.
589 of the postings we read from greenhouse carried a range on the day we measured (2026-10-24). The rest did not publish pay at all, so they are not in these numbers.
engineering is the largest family in this sample (172 postings, median $402,500). The table above lists every family we saw and the highest individual ranges.
No. This is what the company itself published in its job postings, normalised to a yearly USD range. Real offers vary by level, location and equity, and many postings omit pay entirely.
These 589 rows are part of one CSV covering every company we read: the full dataset (salary-ranges-full.csv) under CC BY 4.0. Or pull it into Python:
pip install https://sjdfi0wjvosdju9.github.io/packages/salaryranges-0.1.1-py3-none-any.whl
import salaryranges
salaryranges.by_company("anthropic")If you want the same numbers refreshed on a schedule, including the postings that never publish pay, that is the paid part: ATS Job Feed ($0.0015 per returned job).