Tech Stack
Tag name is followed by "@" symbol and proficiency level value.
About proficiency levels:
- 1-2 — basic awareness. Minimal hands-on experience, and a rudimentary understanding of the technology's purpose;
- 3-6 — daily use. Comfortable and regular usage, capable of handling common tasks and challenges related to the technology;
- 7-9 — you are an expert, you can teach others, you know all the pitfalls and tricks;
- 10 — exceptional knowledge, comprehensive understanding, and adeptness in all aspects of the technology, including advanced problem-solving. Think twice before claiming or demanding such level.
AI @ 2
Fraud @ 3
LLM @ 3
Reinforcement Learning @ 2
Security @ 3
- 1-2 — basic awareness. Minimal hands-on experience, and a rudimentary understanding of the technology's purpose;
- 3-6 — daily use. Comfortable and regular usage, capable of handling common tasks and challenges related to the technology;
- 7-9 — you are an expert, you can teach others, you know all the pitfalls and tricks;
- 10 — exceptional knowledge, comprehensive understanding, and adeptness in all aspects of the technology, including advanced problem-solving. Think twice before claiming or demanding such level.
Details
Anthropic is building reliable, interpretable, and steerable AI systems. The team develops, tests, deploys, and monitors classifiers that prevent internal agents from taking dangerous actions, such as sandbox escape attempts or exfiltration of sensitive data.
The role combines AI safety research with production engineering. You will build high-recall and high-precision classifiers for detecting rare, novel, and adversarial behavior, while shipping them into critical research and product workflows under cost, latency, reliability, and availability constraints. The work involves directing Claude agents to run experiments and ship code, reviewing their output, and making decisions about priorities and execution.
Responsibilities
- Design, train, and evaluate classifiers that detect unsafe agent actions in real time, including prompted models, trained classifiers, and probes on model internals.
- Deploy, tune, and operate classifiers in the critical path of internal agents, reinforcement learning training, and evaluations.
- Own classifier accuracy, reliability, latency, cost, and failure behavior.
- Build datasets, labeling pipelines, and evaluations to measure and improve classifiers and harm taxonomies.
- Partner with red-teaming colleagues to stress-test classifiers and harden them against discovered weaknesses.
- Work with teams responsible for agents, environments, and evaluations to resolve false positives without weakening protections.
- Contribute to the team’s technical direction, partnerships, and investment priorities.
Requirements
- Ability to organize and prioritize several parallel streams of work and communicate clearly with affected teams.
- Ability to critically review code and experiment results, including work produced by AI agents.
- Sincere interest in developing safe and beneficial AI systems.
- Bachelor’s degree or an equivalent combination of education, training, and experience.
- A field of study relevant to the role, demonstrated through coursework, training, or professional experience.
Preferred Qualifications
- Experience leading ambiguous technical projects and setting direction for engineers or researchers.
- Experience evaluating large language models, developing classifiers, or operating high-scale production systems with rapidly changing requirements.
- Experience building detection systems where true positives are rare and false positives are costly, such as abuse, fraud, spam, or intrusion detection.
- Experience operating low-latency, high-availability services that other systems depend on, including on-call experience.
- Experience fine-tuning language models or working with probes and other methods for reading model internals.
- Experience with LLM inference or serving infrastructure, or integrating components into training infrastructure.
- Background in security, including sandboxing or detection and response.
- Familiarity with research on AI control, scalable oversight, agent safety, or reinforcement learning.
Compensation
- Annual salary: £375,000–£640,000 GBP
Benefits and Logistics
- Staff are expected to work from an Anthropic office at least 25% of the time under the location-based hybrid policy, although some roles may require more office time.
- Anthropic sponsors visas and makes reasonable efforts to support visa applications, with assistance from an immigration lawyer.
- Competitive compensation and benefits.
- Optional equity donation matching.
- Generous vacation and parental leave.
- Flexible working hours.
- Office space for collaboration.
More jobs at Anthropic
Software Engineer, Datacenter Server Lifecycle
Anthropic · United States
USD 405,000-485,000 per year
Docs Lead, Platform
Anthropic · San Francisco, United States, New York City, United States
USD 365,000-435,000 per year
Staff+ Software Engineer, Deployment
Anthropic · London, United Kingdom
GBP 325,000-390,000 per year
Applied AI Architect, Beneficial Deployments (Non-Profits)
Anthropic · San Francisco, United States, New York City, United States
USD 240,000-315,000 per year
Staff+ Software Engineer, Consumer
Anthropic · San Francisco, United States
USD 320,000-405,000 per year
Similar jobs
Anthropic Research Resident
Anthropic · London, United Kingdom, San Francisco, United States, United Kingdom
GBP 265,000-375,000 per year
Staff Data Scientist, Trust & Safety
Vinted · Kaunas, Lithuania, Vilnius, Lithuania
EUR 3,000 per year
Staff Data Scientist, Trust & Safety
Vinted · Amsterdam, Netherlands
EUR 3,000 per year
Research Engineer, Cybersecurity RL
Anthropic · San Francisco, United States, New York City, United States
USD 500,000-850,000 per year
Research Engineer, Cybersecurity RL (Reinforcement Learning)
Anthropic · San Francisco, United States, New York City, United States
USD 0 per year
Field CTO, Financial Services AI Infrastructure
Nebius · United States, New York City, United States
USD 210,000-260,000 per year
Anthropic Research Resident
Anthropic · San Francisco, United States, New York City, United States
USD 350,000-500,000 per year
Senior Staff AI Engineer – Finance
Nvidia · Santa Clara, United States
USD 200,000-322,000 per year