ML Security Engineer

Main mission

Protects models against attacks: poisoning, extraction, adversarial.

5 key responsibilities

• Test model resistance to attacks: adversarial, poisoning, extraction.
• Secure training pipelines and model weights.
• Detect malicious uses and attempts to bypass security.
• Define security controls specific to ML systems.
• Collaborate with traditional cybersecurity and ML teams.

Key skills

Adversarial ML, application security, applied cryptography, red teaming, MLOps.

What's expected

Models as well defended as the systems they control: a new and critical discipline.

Career paths

AI Red Teamer, AI Security Architect, AI-focused CISO.