Senior ML Researcher, Deepfake Detection [hosted by University of Twente]

Sensity AI is a forensic-grade platform for deepfake detection and media analysis across images, video, and audio. Combining deep learning with digital forensics, it enables law enforcement, forensic investigators, and intelligence agencies to verify digital evidence, investigate AI-driven threats, and generate explainable, court-ready reports at scale.

Founded in 2018 by AI researchers, Sensity is the world's first company specialised in the detection of AI-generated content. We are backed by Auriga Cyber Ventures, Betaworks Ventures, Amadeus Capital, and the European Innovation Council, other than by Visa Everywhere and Google News Initiative. Sensity is featured by the likes of the New York Times, the Financial Times and Bloomberg.

This position is part of an EIC-funded project carried out in collaboration with the University of Twente (Enschede). You will be hired by a fixed term contract by the University and perform activities together with Sensity's Machine Learning team. You will be supervised by Prof. Marcello Carioni, University of Twente.

This role is for for candidate who are resident in the Netherlands, or are willing to relocate.

The role

• Research, design and train detection models for manipulated and synthetic image, video and audio content
• Keep detectors robust as new generators appear, and under real-world conditions such as compression, re-encoding, cropping and adversarial manipulation
• Build and curate datasets and evaluation benchmarks
• Take models from prototype to production together with our engineering team
• Publish at top venues and contribute to the project research output

What you bring

• PhD in Computer Science, AI or a related field, or an MSc with 3+ years of relevant research or industry experience, with a focus on computer vision, audio or machine learning
• Hands-on experience with deep generative models (diffusion models, GANs, VAEs, autoregressive models)
• Strong PyTorch skills. Experience with the Hugging Face ecosystem is a plus
• Clean, well-structured, reusable code
• Clear communication and a collaborative way of working

Nice to have

• Publications at CVPR, ICCV, ECCV, NeurIPS, ICML, ICLR or similar
• Experience in media forensics, audio deepfakes or adversarial robustness

What we offer

• Research problems with real-world impact, working alongside people who were early to deepfake detection
• A small, international team where your work goes straight into the product
• Flexible hours and remote work