Multimodal ML Engineer for Ad Content & Risk

TikTok Inc.; San Jose (CA); On-site; USD 162,000 - 317,000; Full time

Job summary

TikTok Inc. in San Jose is seeking a Research Engineer for its Business Integrity team. You will help build scalable ML models for ad content understanding and advertiser value, collaborating with global teams to ensure high-quality user experiences.

You will advance multimodal models, work with LLM/MLLM and AIGC technologies for e-commerce and short-video content, and contribute to interpretable decision-making and robust content safety across platforms.

Qualifications

• 1+ year of algorithm R&D/project implementation experience in content understanding or AIGC.
• Solid ML/DL foundation with understanding of MLLM/LLM, CV, NLP, multimodal fusion and Agent tech; problem solving abilities."],
• job_responsibilities_description':['Focus on ad content understanding and security, conduct algorithm R&D for large model implementation, and apply LLM/MLLM and other AIGC technologies to e-commerce and short-video ad content understanding to build a next-gen large model-based commercial intelligence review system.','Deepen multimodal understanding tech for text, audio, video, live streaming, etc., optimize model decision-making for high-accuracy autonomous risk judgment, and implement interpretable CoT generation for traceable model decisions.','Explore RL/Agent applications in multimodal review scenarios, track AIGC cutting-edge trends, and deliver algorithm innovation and engineering implementation tailored to commercial business needs.','Develop dedicated multimodal content understanding models to empower ad intent recognition, intelligent rule retrieval and accurate risk judgment, enhance advertisers' creation experience, reduce non-compliant content non-detection risks, and prote
• CoT_job_summary_short
• COMPANY NAME: TikTok Inc.KEY POINTS: ad content understanding, scalable ML models, multimodal review
• job_summary_short
• <p>TikTok Inc. in San Jose is seeking a Research Engineer for its Business Integrity team. You will help build scalable ML models for ad content understanding and advertiser value, collaborating with global teams to ensure high-quality user experiences.</p><p>You will advance multimodal models, work with LLM/MLLM and AIGC technologies for e-commerce and short-video content, and contribute to interpretable decision-making and robust content safety across platforms.</p>
• contract_type1??
• location_type
• on site
• remote_scope":null,

Responsibilities

• Focus on ad content understanding and security, conduct algorithm R&D for large model implementation, and apply LLM/MLLM and other AIGC technologies to e-commerce and short-video ad content understanding to build a next-gen large model-based commercial intelligence review system.
• Deepen multimodal understanding tech for text, audio, video, live streaming, etc., optimize model decision-making for high-accuracy autonomous risk judgment, and implement interpretable CoT generation for traceable model decisions.
• Explore RL/Agent applications in multimodal review scenarios, track AIGC cutting-edge trends, and deliver algorithm innovation and engineering implementation tailored to commercial business needs.
• Develop dedicated multimodal content understanding models to empower ad intent recognition, intelligent rule retrieval and accurate risk judgment, enhance advertisers' creation experience, reduce non-compliant content non-detection risks, and protect ad ecosystem security.

Skills

ML/DL theory

LLM/MLLM

multimodal fusion

problem solving

Python/C++

Education

Bachelor's degree or above in CS/AI/math

Tools

PyTorch

TensorFlow

Job description

TikTok Inc. in San Jose is seeking a Research Engineer for its Business Integrity team. You will help build scalable ML models for ad content understanding and advertiser value, collaborating with global teams to ensure high-quality user experiences.

You will advance multimodal models, work with LLM/MLLM and AIGC technologies for e-commerce and short-video content, and contribute to interpretable decision-making and robust content safety across platforms.