Data Scientist - Computer Vision / AI Manipulation Detection, Risk AI

About the Role:  As the digital-asset industry continues to evolve, so does the sophistication of identity fraud, deepfake attacks and AI-generated content. To stay ahead of emerging threats, Binance Risk is investing heavily in advanced computer vision and AI-driven identity verification technologies to protect our users and strengthen platform security.

We are looking for an experienced Data Scientist with strong expertise in computer vision, particularly in deepfake detection, face anti-spoofing, facial animation detection, video manipulation detection and video analysis, to join our Risk AI team.

You will research, develop and deploy advanced AI models to detect face spoofing, deepfakes, manipulated videos and other sophisticated identity attacks. You will work on challenging real-world problems involving facial recognition, liveness detection, video understanding and generative AI, translating cutting-edge research into scalable production solutions.

What You’ll Do:

• Research, design and develop advanced computer vision and deep learning models to detect sophisticated identity attacks, including face spoofing, deepfakes, face swapping, face animation, AI-generated faces and videos, and other emerging manipulation techniques.
• Explore state-of-the-art image and video modeling techniques, including Vision Transformers (ViTs), multimodal models and vision-language models (VLMs), to improve fraud detection capabilities.
• Explore how LLMs, agents, graph learning and anomaly detection can be applied to risk scenarios, including automated investigation, risk reasoning and real-time monitoring.
• Build and optimize end-to-end machine learning pipelines, handling large-scale datasets, noisy labels, model training, evaluation and deployment, while improving model accuracy, robustness and inference efficiency.
• Collaborate with KYC, backend engineering, risk operations and strategy teams to deploy scalable AI solutions into production systems.
• Stay current with advances in computer vision, video understanding and generative AI, proactively identifying emerging fraud patterns and translating research breakthroughs into effective risk control solutions.

What We’re Looking For:

• Master's degree or above in Computer Science, Artificial Intelligence, Machine Learning, Computer Vision or a related field, or equivalent practical experience.
• Strong expertise in computer vision and deep learning, particularly in face recognition, face anti-spoofing, deepfake detection, facial animation detection or video understanding.
• Solid understanding of modern deep learning architectures, including CNNs, Vision Transformers (ViTs), multimodal models and video-based models.
• Proficiency in Python and deep learning frameworks such as PyTorch, with hands-on experience in model training, fine-tuning and evaluation.
• Experience handling large-scale image and video datasets, including data preprocessing, noisy labels, data augmentation and model performance analysis.
• Experience developing and deploying machine learning models in production, with a focus on accuracy, robustness, scalability and inference efficiency.
• Strong problem-solving, communication and collaboration skills, with the ability to independently drive complex AI projects.

Nice to Haves:

• Extensive experience in biometric authentication, liveness detection, deepfake detection, video forgery detection or adversarial machine learning.
• Hands-on experience with LLMs, VLMs, agents, knowledge distillation, self-supervised learning or generative AI technologies.
• Experience with large-scale ML infrastructure, distributed training, GPU optimization (TensorRT, ONNX, CUDA) or cloud platforms such as AWS SageMaker.
• Publications at top-tier AI conferences (CVPR, ICCV, ECCV, NeurIPS, ICML, ICLR), notable open-source contributions or prior experience in identity verification, fraud detection or AI security.