AI Red Teamer (Remote)
Handshake · Remote · Remote (USA) ·
COMPENSATION
€•• – ••k
Listed on the employer's job board (Ashby)
- Fully remote (USA)
- W-2 hourly employment
- 40 hours per week
- AI safety focus
- No California residents
Join Handshake AI as a generalist Red Teamer focused on improving model robustness and safety. You will creatively probe LLMs to uncover issues like bias, hallucinations, and guardrail failures, supporting leading research labs in responsible AI development.
What you'll do
- Design creative prompts to test AI guardrails across risk categories
- Use jailbreak and evasion techniques to bypass safety filters
- Explore edge cases to trigger harmful or incorrect model outputs
- Evaluate responses using structured harm taxonomies and severity rubrics
- Document experiments clearly including methods and findings
- Review and refine adversarial prompts created by team members
- Collaborate with engineers and researchers to strengthen defenses
- Stay updated on new attack methods and model behaviors
What you'll need
- Strong hands-on experience with multiple LLMs like ChatGPT and Claude
- Intuition for crafting adversarial prompts and jailbreak techniques
- Creative problem-solving skills with an adversarial mindset
- Clear written communication and strong ethical judgment
- Self-directed work style comfortable with feedback and failure
- Ability to professionally handle disturbing content regularly
Nice to have
- Familiarity with Python or other scripting languages
- Experience with LLM APIs or evaluation tooling
- Comfort with structured data annotation and rubric scoring
- Background in trust and safety, QA, or security research
- Expertise in high-risk domains like cybersecurity or medicine
How you'll work
- Remote work within the USA
- W-2 hourly employment via TCWGlobal
- 40 hours per week schedule
- Candidates in California are not eligible