Note you must apply at this link in order to be considered: https://expert-hub.sepalai.com/application?applicationFormId=a3e099bb-4b46-4c57-b46c-20e51ca12bf7&outreachCampaignId=3c4a0b58-05b7-406f-8da8-6e9f7a3f90af Sepal AI partners with top AI research labs (OpenAI, Anthropic, Google DeepMind) to determine how dangerous AI models are. We need hackers, red-teamers, and CTF veterans to work with the newest LLM coding models to understand what they're capable of and to hack our CTFs. 🧠 What You'll Do - Design adversarial scenarios that probe AI assistants for injection, privilege-escalation, and data-exfiltration risks. - Execute red-team engagements against AI-enabled workflows (web, cloud, and SaaS). - Craft realistic exploit chains and payloads a real attacker might use, then measure whether the model blocks or facilitates the attack. - Build scoring rubrics, attack trees, and reproducible test harnesses to grade model resilience. - Collaborate with AI researchers to iterate on defenses, patch vulnerabilities, and improve sandboxing & policy enforcement. ✅ Who You Are - OSCP (or comparable) certification and 3+ years in professional red teaming or offensive security. - Proven CTF track record; comfortable with web, cloud, and application exploits. - Fluent in Python/Bash tooling, custom payload development, and exploit automation. - Strong grasp of common enterprise stacks (OAuth, SAML, REST, SQL, cloud IAM). - Passionate about AI safety and eager to keep cutting-edge models from becoming attack surfaces. 💸 Pay: $40–90/hr 🌍 Remote ⏱ Flexible hours (project-based, async)