AI Safety Research Scientist

Responsibilities:

• Lead the research, design and development of highly trustworthy, robust and reliable, high performance, efficient detection solutions, generative service and Safe Agentic AI systems.
• Contribute to the development of Online Safety Research Roadmap to advance AI-based detection solution and ensure the safety of AI systems of classification, generative AI and agentic AI systems.
• Provide technology insights and research proposal for advancing the state-of-the-art research and development.
• Design and evaluate solution architecture and design, assess the competitiveness of the technology.
• Conduct hands-on research and experiments in collaboration with team members.

Key Requirements:

• PhD in Computer Science, Deep Learning, Machine Learning, Mathematics or other related fields.
• 8+ years of experience, focused on the research in AI and/or Security field with a strong track record and high motivation to the Trust & Safety domain.
• Proven experience in designing and building efficient, high-performance LLMs/VLMs and agentic AI / reasoning systems, including model architecture, optimization and large-scale deployment.
• Successful experience in advanced AI in aligning AI with human values and expectations such as RLHF, adversarial training, neural network editing, formal logic & game theory, capabilities to detect and understand risks, intent modeling and reasoning, robustness, capabilities of understanding local cultures and explainable AI.
• Experience in security threat or abuse detection, deep fake detection, fairness, transparency, robust and explainable AI is a plus.
• Experience implementing solution that comply with EU online safety regulations and privacy protection laws (e.g., Digital Service Act, GDPR, AI Act) is a plus.

Desirable Skills:

• Pioneering novel methods and neural networks that revolutionized machine learning or the AI field, or revolutionized the industry, is a big bonus.
• International awards in the field of AI/ML/CV and highly recognized by experts of the same field.
• Strong knowledge and experience in LLM/VLM pre-training models, neural network architecture and algorithm design.

• Strong publication record in top conferences e.g., AAAI/ACL/CVPR/ICCV/EMNLP/NAACL