AI Security Researcher / Research Engineer
Check Your Resume Before You Apply
Upload your resume to compare your skills and keywords with this job's actual requirements.
✓ Skills That Match This Job
✗ Not Found In Your Resume
⚠️ Missing Important Resume Keywords?
Your resume may need to be optimised before you apply. Improve your resume with an ATS-friendly resume builder.
What Similar Candidates Looked Like
Illustrative only — replace with real hiring insights for this role if you have them.
Top Skills
Previously Worked As
Overview
Brave builds a privacy-first web browser used by over 120 million people, along with a private search engine, a private AI assistant, and an opt-in private ad network. This role focuses on the security risks that come with building AI systems — especially LLM-powered web agents and chatbots. You'll design secure-by-design architectures, audit Brave's LLM-powered web agent and chatbot, and help set the company's direction on identifying, measuring, and reducing AI security risk.
Responsibilities
- Design and execute adversarial tests to uncover failure modes in LLM-powered web agents and chatbots (e.g., prompt injection, tool abuse, data exfiltration).
- Analyze model behavior—including internal chain-of-thought outputs—to forecast and prevent security risks in frontier models.
- Augment web agents with robust security principles, including information flow control, privilege control, and contextual security.
- Develop and prototype security-focused training and fine-tuning techniques for AI models.
- Help define forward-looking security strategies and policies for building and responsibly releasing AI web agents while advocating for user protection.
- Work alongside engineers and researchers in privacy, machine learning, and cryptography to translate findings into production guardrails, tests, and product enhancements.
Qualifications
Required / Minimum Qualifications
- Secure AI Expertise: Demonstrated experience working on Secure AI, specifically prompt injection attacks and building security guardrails.
- Research & Publication Record: Strong publication record in AI security, ML security, or closely related areas.
- Product Shipping Track Record: Proven experience shipping products and features inside a fast-moving, high-execution environment.
- Programming & ML Frameworks: Strong coding skills in Python and hands-on familiarity with machine learning frameworks like PyTorch or JAX.
- LLM & Web Agent Mastery: Deep familiarity with large language models and web agents—how they function, operate, and are trained.
Additional / Preferred Qualifications
- Top-Tier Doctorate: Ph.D. in Computer Science from a top-tier school with a strong publication record in premier systems, performance, or machine learning venues.
- AI Systems Knowledge: End-to-end understanding of AI system architectures, deployment pipelines, and security evaluation frameworks.
- Advanced Privacy & Cryptography: Hands-on experience or familiarity with Confidential Computing and ZKP (Zero-Knowledge Proofs) or SMPC (Secure Multi-Party Computation) protocols.
⚠️ Work Authorization
WORK_AUTHORIZATION_TEXT: Not specified in the job listing — no sponsorship or country-eligibility requirement is stated.
🎯 Need Expert Help With Your Job Search?
From resume writing to LinkedIn profile optimization and interview coaching — get expert freelance help to strengthen your job application on Fiverr.
GET EXPERT HELP ON FIVERR →Ready to Apply?
Apply directly through the official Brave career page.
APPLY ON OFFICIAL EMPLOYER CAREER PAGE →Frequently Asked Questions
This role focuses on finding and mitigating security vulnerabilities in LLM-powered systems, primarily Brave's AI assistant (Leo) and autonomous web agents. Key areas include defending against indirect prompt injection, preventing data exfiltration, controlling tool abuse, and auditing model inner reasoning steps.
Brave does not specify a rigid minimum number of years (e.g., "5+ years"). Instead, the focus is on demonstrated expertise in Secure AI, prompt injection defense, Python, and ML frameworks (PyTorch/JAX), along with a track record of shipping features or publishing research.
The role is based out of London, England, with options for hybrid or remote work. Exceptional candidates located elsewhere in the UK or across Europe will also be considered.
The expected US salary range for this position is $180,000 to $225,000 per year, depending on experience, plus benefits and professional development opportunities.
A Ph.D. in Computer Science (focusing on AI security, machine learning, or system safety) is strongly preferred. Candidates with equivalent industry experience and a strong track record of shipping production AI safety tools or publishing in top-tier security/ML conferences are encouraged to apply.
You will write production code primarily in Python and build or modify security guardrails using machine learning frameworks such as PyTorch or JAX. Familiarity with LLM evaluation pipelines and agentic architecture is required.
Brave builds privacy-first, client-side, and agentic AI products directly integrated into a web browser used by over 120 million people. Because web agents interact dynamically with untrusted web content, security research here directly impacts user data protection in real-world browsing environments.
You May Also Like
Related Video
Share This Job
Joblyst2025 is an open-access job information and career resources website. Job details are sourced from publicly available employer career pages. Candidates should verify all job details, eligibility requirements and application status on the official employer website before applying.
Some links on this page may be affiliate links. If you purchase a product or service through an affiliate link, Joblyst2025 may receive a commission at no additional cost to you.
