How to apply for Trusted Red Team Network

About Syntony

Syntony is a niche consultancy that directly supports frontier AI organizations and public institutions in safety research and adversarial evaluation. Their focus on translating technical findings into governance and risk advisory makes them a unique player in AI safety, offering contract-based work that is both intellectually rigorous and practically impactful.

About the role

As a Trusted Red Team Network member, you will conduct project-based adversarial evaluations to identify vulnerabilities in AI systems, collaborating across disciplines to produce actionable controls and oversight artifacts. Your work will directly inform mitigation decisions and governance design, making this a high-impact role for those passionate about AI safety.

A typical day

You'll start by reviewing the threat model and scope for a new engagement, then spend the day designing and executing adversarial tests (e.g., prompt injection attacks, policy boundary probes). You'll document findings in a structured report, collaborate with a policy analyst to refine actionable controls, and attend a brief sync with the client to present initial results.

Who Syntony is looking for

• Expert in ML security with hands-on experience in adversarial attacks and defense mechanisms for large language models or other AI systems.
• Proficient in prompt engineering techniques such as jailbreaking, prompt injection, and evasion attacks, with a track record of discovering novel vulnerabilities.
• Strong systems thinking and policy analysis skills, able to translate technical findings into clear, actionable recommendations for non-technical stakeholders.
• Comfortable working in scoped, project-based engagements with defined threat models and strict evidence requirements, often under tight deadlines.

Tips for this application

• Tailor your resume to highlight specific adversarial evaluation projects you've led, including the threat model, attack vectors tested, and resulting mitigations.
• Include a portfolio or examples of technical write-ups from past red teaming engagements, especially those involving ML security or prompt engineering.
• Demonstrate interdisciplinary collaboration by mentioning experience working with policy experts, engineers, or governance teams on AI safety projects.
• Explicitly address how you translate technical findings into actionable controls or oversight artifacts, as this is a key requirement.
• Research Syntony's published work (e.g., on their website or GitHub) and reference specific methodologies or frameworks they use in your cover letter.

What to cover in your cover letter

['Emphasize your expertise in one or more of the specified areas (ML security, prompt engineering, policy analysis, geopolitical forecasting, or systems thinking) with concrete examples.', 'Highlight your ability to work in scoped, project-based engagements and produce clear, actionable recommendations from technical findings.', "Show your understanding of the importance of adversarial evaluation in informing governance and risk decisions, tying it to Syntony's mission.", "Mention any experience with frontier AI organizations or public institutions, as this aligns with Syntony's client base."]

Research before applying

• Read Syntony's website and any published whitepapers or reports on their research methodologies and past projects.
• Look into their client base (frontier AI organizations, public institutions) to understand the types of systems you might evaluate.
• Review recent adversarial evaluation trends in AI safety, such as new attack vectors or evaluation frameworks from organizations like ARC Evals or METR.
• Check if Syntony has a blog or GitHub repository with code examples or case studies that can inform your interview preparation.

Likely interview topics

Based on the job description, expect questions about:

• Describe a past adversarial evaluation project: the threat model, attack methods, key findings, and how you translated those into controls.
• How do you approach prompt engineering for a novel AI system you've never seen before? Walk through your methodology.
• Given a specific AI system (e.g., a chatbot), design a red teaming plan with defined threat models and evidence requirements.
• How do you balance depth vs. breadth in a scoped engagement when you have limited time? Provide an example.
• Explain a complex technical vulnerability to a non-technical policymaker. What key points would you emphasize?

Common mistakes to avoid

• Submitting a generic resume without specific examples of adversarial evaluation or red teaming in AI/ML contexts.
• Focusing only on technical skills without demonstrating the ability to communicate findings to non-technical stakeholders.
• Failing to show understanding of scoped engagements and evidence-based requirements, instead treating the role as open-ended research.