Red Teaming Fellowship

10a Labs; San Francisco (CA); Hybrid; USD 34,000 - 45,000; Full time

Benefits offered by this job

Flexible start/end dates

Remote work (continental U.S.)

Flexible schedule, up to 20 hours per‑

Hourly pay commensurate with exp

Job summary

10a Labs is seeking a Red Teaming Fellow to help evaluate the safety, security, and reliability of advanced AI systems. You will work alongside researchers and engineers to identify vulnerabilities and generate insights that guide safer deployment of frontier models and AI-enabled systems.

This hands-on fellowship sits at the intersection of AI, security, and adversarial testing, focusing on real-world evaluations and crafting tests that stress model behavior under challenging conditions.

Qualifications

• Pursuing or recently completed a degree in Computer Science, Cybersecurity, Engineering, Data Science, Mathematics, Political Science, International Relations, or a related field
• Strong interest in AI safety, cybersecurity, red teaming, or adversarial testing
• Familiarity with AI abuse including jailbreaking, system prompt extraction, indirect prompt injection, data exfiltration, etc.
• Strong analytical, research, problem‑solving, and communication skills
• Ability to think creatively from an adversarial perspective
• Ability to work independently and collaboratively in a remote environment
• Experience with Python, Bash, large language models, AI systems, cybersecurity, or research methodologies preferred
• Proficiency in a language other than English preferred

Responsibilities

• Conduct adversarial testing of AI systems across safety, security, and misuse scenarios
• Design and execute red teaming exercises against models, agents, and AI-enabled applications
• Develop prompts, attack strategies, and test cases to identify vulnerabilities and failure modes
• Analyze model behavior and document findings in clear, actionable reports
• Support the development of evaluation frameworks, taxonomies, and testing methodologies
• Research emerging AI threats, attack techniques, and risk trends
• Collaborate with engineers, researchers, and subject-matter experts to investigate emerging risks and threat vectors
• Develop front‑end dashboards and other visualizations

Skills

Adversarial testing

Analytical skills

Communication skills

Independent work

Education

Computer Science

Cybersecurity

Engineering

Data Science

Mathematics

Political Science

International Relations

Tools

Python

Bash

LLMs

Job description

About 10a Labs:

10a Labs is the safety and threat‑intelligence layer trusted by frontier AI labs, AI unicorns, Fortune 10 companies, and leading global technology platforms. Our adversarial red teaming, model evaluations, and intelligence collection enable engineering, safety, and security teams to stay ahead of evolving threats and deploy AI systems safely.

Location & Travel Expectations

We are primarily seeking candidates based in the Washington, DC, New York City, San Francisco Bay Area, and Denver metropolitan areas. Candidates hired in these locations will be expected to work from a 10a Labs office at least two days per week. All new hires will participate in in-person onboarding in Washington, DC, New York City, or San Francisco during their first week. Occasional travel may also be required for interviews, in-person meetings, and other business needs.

About the role

As a Red Teaming Fellow, you will help evaluate the safety, security, and reliability of advanced AI systems. Fellows work alongside researchers, engineers, and subject-matter experts to identify vulnerabilities, uncover failure modes, and generate insights that help organizations deploy AI systems more safely.

This is a hands‑on fellowship at the intersection of AI, security, and adversarial testing. Fellows will contribute to real‑world evaluations of frontier models and AI-enabled systems, helping design and execute tests that probe how models behave under challenging or unexpected conditions.

In This Role, You Will

• Conduct adversarial testing of AI systems across safety, security, and misuse scenarios
• Design and execute red teaming exercises against models, agents, and AI-enabled applications
• Develop prompts, attack strategies, and test cases to identify vulnerabilities and failure modes
• Analyze model behavior and document findings in clear, actionable reports
• Support the development of evaluation frameworks, taxonomies, and testing methodologies
• Research emerging AI threats, attack techniques, and risk trends
• Collaborate with engineers, researchers, and subject-matter experts to investigate emerging risks and threat vectors
• Develop front‑end dashboards and other visualizations

Qualifications

• Pursuing or recently completed a degree in Computer Science, Cybersecurity, Engineering, Data Science, Mathematics, Political Science, International Relations, or a related field
• Strong interest in AI safety, cybersecurity, red teaming, or adversarial testing
• Familiarity with AI abuse including jailbreaking, system prompt extraction, indirect prompt injection, data exfiltration, etc.
• Strong analytical, research, problem‑solving, and communication skills
• Ability to think creatively from an adversarial perspective
• Ability to work independently and collaboratively in a remote environment
• Experience with Python, Bash, large language models, AI systems, cybersecurity, or research methodologies preferred
• Proficiency in a language other than English preferred

Benefits

• Flexible start / end dates
• Remote work (based in the continental U.S.)
• Flexible schedule, up to 20 hours per week (negotiable)
• Hourly pay commensurate with experience and qualifications
• $25 per hour for undergraduate students
• $32.50 per hour for graduate students