Senior Multimodal ML Scientist — Foundation Models

TikTok Inc.; San Jose (CA); On-site; USD 218,000 - 388,000; Full time

Benefits offered by this job

Medical insurance

Dental insurance

Vision insurance

401(k) match

Parental leave

Paid time off

Job summary

TikTok Inc. is seeking a research-focused engineer to lead design, training, and deployment of foundation models (LLM/VLM) to support content safety across text, image, and video modalities.

You will work with world‑class researchers and ML engineers to build scalable systems that detect abuse and support compliant moderation on a global platform. You will tackle multilingual and low-resource scenarios, optimize large-scale training pipelines, and help align models with evolving safety policies

Qualifications

• 2+ years of research or development with large models.
• Strong problem-solving with track record of technical breakthroughs and business impact.
• Self-driven with strong learning and communication skills; up-to-date with multimodal large models.
• Team player; open to new tech and continuous technical evolution.

Responsibilities

• Lead the design, training, and deployment of foundation models (LLM/VLM) for safety tasks across modalities.
• Improve zero-shot and few-shot generalization in multilingual multimodal safety domains.
• Design and maintain a multimodal safety annotation framework and RLHF strategies.
• Collaborate with infra, data, and platform teams to optimize large-scale training pipelines and model serving.
• Partner with product, policy, and recommendation teams to integrate safety models into moderation workflows.

Skills

Large models experience

Multimodal ML

Zero-shot / few-shot generalization

RLHF knowledge

Python / deep learning

Tools

Python

Job description

TikTok Inc. is seeking a research-focused engineer to lead design, training, and deployment of foundation models (LLM/VLM) to support content safety across text, image, and video modalities.

You will work with world‑class researchers and ML engineers to build scalable systems that detect abuse and support compliant moderation on a global platform. You will tackle multilingual and low-resource scenarios, optimize large-scale training pipelines, and help align models with evolving safety policies