Responsabilidades
- Crafting prompts and scenarios to test model guardrails, exploring creative ways to bypass restrictions, and systematically documenting outcomes.
- Think like an adversary to uncover weaknesses, while collaborating with engineers and safety researchers to share findings and improve system defenses.
Requisitos
- Deep Understanding of Generative AI and main models, including their underlying architectures, training processes, and potential failure modes. This includes knowledge of concepts like prompt engineering, fine-tuning, and reinforcement learning with human feedback (RLHF)
- Cybersecurity & Threat Modeling: Experience in cybersecurity principles, including threat modeling, vulnerability assessment, and penetration testing. Ability to identify attack vectors, simulate real-world threats, and understand the potential impact of an attack.
- Data Analysis & NLP: Strong analytical skills to dissect model outputs, identify subtle biases or factual errors, and recognize patterns in how the model responds to different inputs. A background in Natural Language Processing (NLP) would be highly beneficial.
- Ethical Hacking Mindset: A commitment to using her/his skills for defensive and security-focused purposes, adhering to a strict ethical code, and understanding the importance of responsible disclosure.
- You hold a Bachelor's or Master’s Degree in Computer Science, Software Engineering, Cybersecurity, Digital Forensics or other related fields.
- Your level of English is advanced (C1) or above
- Adversarial thinking
- Knowledge of vulnerabilities, common model vulnerabilities (prompt injection, prompt-history leakage, data exfiltration via RAG).
- Experience in AI/ML security, evaluation, and red teaming, particularly with LLMs, AI agents, and RAG pipelines.
- You are ready to learn new methods, able to switch between tasks and topics quickly and sometimes work with challenging, complex guidelines.
- Proficient in scripting and automation using Python, Bash, or PowerShell.
- Familiar with AI red-teaming frameworks such as garak or PyRIT.
Deseable
- Physical-world adversarial testing
- Experienced with containerization and CI/CD security tools, especially Docker.
- Proficient in offensive exploitation and exploit development.
- Skilled in reverse engineering using tools like Ghidra or equivalents.
- Expertise in network and application security, including web application security.
- Knowledge of operating system security concepts such as Linux privilege escalation and Windows internals.
- Familiar with secure coding practices for full-stack development.
Beneficios
- Get paid for your expertise, with rates that can go up to $55/hour depending on your skills, experience, and project needs.
- Take part in a part-time, remote, freelance project that fits around your primary professional or academic commitments.
- Work on advanced AI projects and gain valuable experience that enhances your portfolio.
- Influence how future AI models understand and communicate in your field of expertise.
Información adicional
- Your level of English is advanced (C1) or above
