Anthropic
Staff+ Software Engineer, Account Abuse (Machine Learning)
$320K–$485K
San Francisco, CA +2 · Staff · Full-time · Posted Sep 28, 2026
- Location
- San Francisco, CA
New York City, NY
San Francisco, California, United States - Experience
- Years required correlate with internal job-level requirements; no specific number stated.
- Education
- Bachelor’s degree or an equivalent combination of education, training, and/or experience. Required field of study: a field relevant to the role, demonstrated through coursework, training, or professional experience.
- Visa sponsorship
- Visa sponsorship is available, but not for every role or candidate; the company says it will make every reasonable effort to obtain a visa after an offer.
About the job
Build and operate machine-learning systems to detect and prevent account abuse and fraud, from feature computation through model deployment and monitoring.
What you'll do
- Build and operate feature computation for model training and low-latency online scoring; train, evaluate, and deploy abuse-detection models.
- Automate model development and establish backtesting, shadow deployment, staged rollout, and monitoring for drift and training/serving skew.
- Collaborate with data science, Policy & Enforcement, product, and platform teams to improve labels and integrate model decisions.
What they ask for
- Proficiency in Python and SQL.
- Experience training machine-learning models and deploying them to production.
- Experience building data pipelines with a batch processing engine and workflow scheduler.
- Understanding of point-in-time correctness and training/serving skew, and how to prevent both.
Nice to have
- Experience with feature platforms, stream processing, or production ML with demanding serving requirements.
- Experience with tabular, unsupervised, clustering-based, or graph-based abuse-detection models; integrity, fraud, or abuse detection; scarce or noisy labels; or ML workflow automation.
Pay
USD 320,000–485,000 annual base salary
Tools
Python, SQL, Spark, Beam, Airflow, Chronon, Feast, Tecton
AI focus
Use Claude to accelerate model feature development, training, evaluation, and maintenance; build classical ML systems for abuse detection.
Model-written summary · the original posting is authoritative