Anthropic’s mission is to create reliable, interpretable, and steerable AI systems that are safe and beneficial for users and society. As part of the Safeguards ML team, the ML/Research Engineer, Safeguards will build systems that detect misuse—from individual policy violations to sophisticated coordinated attacks—and develop defenses that keep our products safe as capabilities advance. The role focuses on developing classifiers, monitoring harm signals across multiple exchanges, evaluating safety of agentic products, and conducting research on automated red‑teaming and adversarial robustness.
ML/Research Engineer, Safeguards at Anthropic
Hybrid - San Francisco, CA, USA; New York City, NY, USA
More jobs at AnthropicSalary
USD 350,000 - 500,000
Requirements
Skills
- 4+ years of experience in ML engineering, research engineering, or applied research, in academia or industry
- Proficiency in Python and experience building ML systems
- Comfortable working across the research-to-deployment pipeline, from exploratory experiments to production systems
- Concern about misuse risks of AI systems and desire to mitigate them
- Strong communication skills and ability to explain complex technical concepts to non-technical stakeholders
- Experience with language modeling and transformers
- Experience building classifiers, anomaly detection systems, or behavioral ML
- Experience with adversarial machine learning or red-teaming
- Experience with interpretability or probes
- Experience with reinforcement learning
- Experience building high-performance, large-scale ML systems
Responsibilities
- Develop classifiers to detect misuse and anomalous behavior at scale, including synthetic data pipelines for training and automatic source of representative evaluations
- Build systems to monitor for harms that span multiple exchanges, such as coordinated cyber attacks and influence operations, and develop new methods for aggregating and analyzing signals across contexts
- Evaluate and improve the safety of agentic products—developing both threat models and environments to test for agentic risks, and developing and deploying mitigations for prompt injection attacks
- Conduct research on automated red‑teaming, adversarial robustness, and other research that helps test for or find misuse
Technologies
PythonLanguage modelingTransformersAdversarial machine learningRed‑teamingInterpretabilityReinforcement learningHigh‑performance ML systems
See if your resume is ready for this job
See how our AI can optimize your resume and improve your chances for this role.