Senior AI Safety & Model Validation Lead

TikTok USDS Joint Venture; Washington (District of Columbia); Presencial; USD 168.000 - 394.000; Jornada completa

Descripción de la vacante

TikTok USDS Joint Venture seeks a senior Model Validation Lead to build the capability and team for independent safety and security assurance of generative AI models. The role sets technical direction, defines testing methods, and leads complex validation engagements across internal and external models.

You will test model safety features, collaborate with Penetration Testing, and develop repeatable evaluations to prove safeguards work before deployment and throughout the lifecycle.

Formación

• Bachelor's degree or equivalent in CS, information security, AI/ML, or related field.
• 8+ years in security testing, adversarial testing, AI security, trust and safety, red teaming, or related.
• Hands-on security testing of generative AI systems and model safety.

Responsabilidades

• Lead the technical direction for model validation and set the testing roadmap.
• Build and scale the validation capability with tooling, pipelines, and expert capacity.
• Design and execute adversarial evaluations of safeguards and data exposures.
• Collaborate with security, engineering, and product teams to identify and mitigate risks.
• Lead pre-deployment and continuous testing of models, including broader AI systems.

Conocimientos

Security testing

Generative AI security

Technical leadership

Evaluation design

Security breadth

Python

Automation

Communication

Educación

Bachelor's degree in CS or related field

Descripción del empleo

TikTok USDS Joint Venture seeks a senior Model Validation Lead to build the capability and team for independent safety and security assurance of generative AI models. The role sets technical direction, defines testing methods, and leads complex validation engagements across internal and external models.

You will test model safety features, collaborate with Penetration Testing, and develop repeatable evaluations to prove safeguards work before deployment and throughout the lifecycle.