Latest AI JobsBrowse Jobs
Anthropic logo

Cyber Evaluations Engineer

AnthropicRemote-Friendly, United States; San Francisco, CA | Washington, DCSafeguards (Trust & Safety)
Apply on company site
Salary
Not listed
Job type
Full-time
Work mode
Remote
Posted
1 Sept 2026
Status
Open · checked 8d ago

About this role

The Cyber Evaluations Engineer role at Anthropic involves building and running evaluations to measure cyber-relevant capabilities and safeguard robustness in models, operating remotely or from San Francisco or Washington, DC. This position suits professionals with a background in cybersecurity, machine learning evaluations, and Python programming who want to help develop safe and beneficial artificial intelligence systems.

Role Overview

Anthropic is looking for a Cyber Evaluations Engineer to design new evaluations, conduct per-release robustness testing, and analyze data on prompt bypasses and jailbreaks. This role focuses on understanding where model safeguards succeed and fail while shaping detection architecture in collaboration with policy teams.

Key Responsibilities

Design and execute capability, uplift, and safety evaluations to assess cyber risks in new models.
Perform per-release safeguard robustness testing prior to major model launches.
Analyze evaluation data and present findings clearly to team members and stakeholders.
Design, prototype, and refine detection probes targeted at cyber misuse.
Partner with the cyber policy team to build a layered abuse-detection architecture and measure its coverage and precision over time.
Develop and maintain internal tooling used for running and scoring evaluations.
Translate evaluation findings into actionable safeguard enhancements alongside engineering and policy partners.

Requirements

Experience creating or running evaluations, benchmarks, or test suites for machine learning systems or software under fixed timelines.
Practical cybersecurity experience such as vulnerability research, CTF participation, exploit development, or security research.
Proficiency in Python programming.
Strong communication skills for sharing evaluation results with cross-functional or policy stakeholders.
Minimum education of a Bachelor's degree or an equivalent combination of training, education, and experience in a relevant field.

Nice to Have

Advanced offensive-security or security-research background, including building AI security benchmarks.
Experience analyzing intrusion telemetry, fraud data, jailbreaks, or adversarial prompt bypasses.
Previous experience working onsite with government partners on evaluation or testing engagements.
Familiarity with AI or ML evaluation frameworks and coordinated vulnerability disclosure practices.
Experience testing pre-release software or models under confidentiality constraints.
Background authoring detection content like YARA, Sigma, Suricata, or SIEM rules, or building ML-based abuse detection.
Active secret security clearance or higher, or eligibility to obtain one.

Skills and Tools

Python
Machine Learning Evaluation Frameworks
Cybersecurity Research Tools and Techniques
Detection Content Authoring (e.g., SIEM, YARA, Sigma)

Salary and Benefits

The annual salary range for this position is $300,000 to $405,000 USD. Anthropic provides competitive compensation and benefits, generous parental leave and vacation, optional equity donation matching, flexible working hours, and visa sponsorship when applicable.

About the Company

Anthropic is a public benefit corporation focused on building reliable, interpretable, and steerable AI systems that are safe and beneficial for society. Operating as a single collaborative team on large-scale research efforts, the organization values empirical science and impactful, long-term advancements in artificial intelligence.

How to Apply

Interested candidates should visit the official Anthropic careers page to submit an application for this position.

Skills and tech stack

Python

Questions candidates ask

What is the salary range for the Cyber Evaluations Engineer?

The annual salary range is $300,000 to $405,000 USD.

Where is this job located?

The role is remote-friendly within the United States, with office presences in San Francisco, CA, and Washington, DC, operating under a location-based hybrid policy expecting staff in-office at least 25% of the time.

What programming language is required?

Proficiency in Python is required for this role.

Does Anthropic offer visa sponsorship?

Yes, Anthropic does sponsor visas and retains an immigration lawyer to assist candidates who receive an offer.

What are the core responsibilities of the role?

Responsibilities include designing and running safety and capability evaluations, executing per-release robustness testing, analyzing evaluation results, prototyping detection probes for cyber misuse, and collaborating with policy teams to build abuse-detection architectures.

Ready to apply?

The application is on the official Anthropic careers page.

Apply on company site