Senior ML Engineer — Red-teaming
Posted 72d ago
Posted by Karokan
karokan.ai
€100–180/hr
0 of 3 steps completed · 0%
Submit your resume or equivalent profile to start. We review background, communication quality, and evidence of relevant execution experience before moving forward.
ⓘ Application steps are reused across roles — you never upload your resume or take the same interview twice.
Role Overview
Design adversarial suites for agentic workflows, tool misuse, multilingual jailbreaks, and instruction hijacking. Work directly with safety researchers on frontier models.
Karokan is working with AI labs and deployment teams across Europe to contract high-judgment specialists who can improve model quality through reliable human input. This role is designed for experts who can translate nuance into structured outputs, work with clear evaluation rules, and contribute directly to the performance of production AI systems.
You will operate in a fully remote setting, collaborate asynchronously, and help shape datasets, reviews, and decision layers that support the next generation of AI products. The work is intentionally rigorous: accuracy, consistency, and written reasoning matter more than generic throughput.
Key Responsibilities
- Design and execute adversarial prompt suites against deployed models
- Probe for tool misuse, privilege escalation, and instruction hijacking
- Develop multilingual jailbreak evaluations across French and Spanish
Ideal Qualifications
- Strong Python and ML engineering background
- Demonstrated experience in LLM security or adversarial ML
