AI Risk & Integrity Engineer

TikTok USDS Joint Venture; San Jose (CA); On-site; USD 150,000 - 259,000; Full time

Job summary

TikTok USDS Joint Venture is seeking a PhD-level candidate to shape the PaCI team's risk infrastructure. You will architect algorithms that secure a billion-user ecosystem across domains including accounts, content safety and financial risk.

You will leverage Transformer and GNN approaches on multimodal data, enhance AI-enabled risk detection, and contribute to robust, scalable defenses and governance. Onboarding by year-end is expected.

Qualifications

• PhD or near-completion in CS/Engineering/Math/Statistics or related field.
• Strong ML foundation; experience with Transformer and GNN architectures for large-scale integrity problems.
• Proficient in Python, C++, or Go; familiar with distributed computing (Spark, Flink) and DL frameworks (PyTorch).
• Excellent communication in a global, cross-cultural environment.

Responsibilities

• Lead algorithmic architecture and governance for risk across domains such as account integrity, content safety, and financial risk.
• Build high-precision risk representations using large heterogeneous data.
• Research and counter adversarial threats with self-evolving defense systems.
• Drive integration of Generative AI and Multimodal tech into risk lifecycle; balance precision with automated enforcement.

Skills

Machine Learning

Transformer models

GNN

Distributed computing

Python

C++

Go

PyTorch

Cross-cultural communication

Education

PhD in Computer Science, Engineering, Math, Statistics, or related

Tools

PyTorch

Spark

Flink

Job description

TikTok USDS Joint Venture is seeking a PhD-level candidate to shape the PaCI team's risk infrastructure. You will architect algorithms that secure a billion-user ecosystem across domains including accounts, content safety and financial risk.

You will leverage Transformer and GNN approaches on multimodal data, enhance AI-enabled risk detection, and contribute to robust, scalable defenses and governance. Onboarding by year-end is expected.