AI Safety Analyst: Model Risk & Data Quality

ByteDance; Singapore; On-site; SGD 150,000 - 190,000; Full time

Job summary

ByteDance invites a proactive AI Safety professional to join the Seed AI Data & Safety team. You will spearhead safety evaluations across LLM, VLM, and multimodal models, collaborating with researchers, data engineers, and product teams to improve model safety and policy alignment.

The role involves designing robust evaluation metrics, driving end-to-end projects, and creating scalable safety workflows to mitigate risk while enabling responsible AI innovation.

Qualifications

• Strong command of English in both written and oral communication.
• Strong analytical skills, with the ability to interpret qualitative and quantitative data and translate them into insights.
• Proven project management abilities, with experience leading cross-functional initiatives in dynamic environments.
• Creative problem-solving mindset, with comfort working under ambiguity and leveraging tools to improve processes.

Responsibilities

• Conduct research on AI safety developments in academia and industry; identify limitations in evaluation paradigms and propose novel testing approaches.
• Design and refine safety evaluation guidelines for multi-model systems; define metrics for safety-related behaviours and alignment with responsible AI principles.
• Analyze safety evaluation results to surface issues from training, fine-tuning, or product integration and translate findings into actionable insights.
• Partner with cross-functional stakeholders to build scalable safety evaluation workflows and feedback loops for continuous model improvement.
• Manage end-to-end project lifecycles including scoping, planning, execution and delivery; allocate resources and coordinate across functions.
• Innovate on automated workflows and methodologies for safety data generation, assessment, and policy research.

Skills

English proficiency

Analytical skills

Project management

Creative problem solving

Education

Bachelor’s degree or higher

Job description

ByteDance invites a proactive AI Safety professional to join the Seed AI Data & Safety team. You will spearhead safety evaluations across LLM, VLM, and multimodal models, collaborating with researchers, data engineers, and product teams to improve model safety and policy alignment.

The role involves designing robust evaluation metrics, driving end-to-end projects, and creating scalable safety workflows to mitigate risk while enabling responsible AI innovation.