Senior Research Scientist
ABOUT THE ROLE
Proving who someone is online has never been harder or more important. You'll research how to detect fraud and verify identities more accurately. Your work will include detecting AI-generated and manipulated images, understanding how models perform across populations, and finding patterns in how attackers behave.
You bring depth in at least one of computer vision, mathematics and statistics, or identity verification and fraud prevention. You test ideas with focused experiments and follow what the results show. You'll work with our ML and engineering teams to turn your findings into defenses for real customers, and you'll get to publish your work.
WHAT YOU’LL DO AT PERSONA
- Lead research on open problems in identity verification and fraud prevention, using focused experiments to test new ideas quickly.
- Develop methods to detect manipulated documents, AI-generated images and emerging fraud attacks.
- Evaluate how reliably our models perform across different populations
- Uncover the patterns in verification data that reveal fraud, and prove they're real signal and not noise.
- Calibrate model confidence and measure fraud events with limited labeled data.
- Partner with ML and engineering to turn research prototypes into production models.
- Publish at top venues such as CVPR, NeurIPS and USENIX Security, and stay active in the research community.
WHAT YOU’LL BRING TO PERSONA
- Deep expertise in one of the following:
- Computer vision: research in image forensics, face analysis, generative models or representation learning.
- Mathematics and statistics: strong foundation in areas such as probability, statistical inference, optimization, or applied mathematics, with experience applying it to empirical research.
- Identity or fraud prevention: Experience researching identity verification, biometrics, fraud detection, out-of-distribution detection, adversarial behavior, or closely related security problems.
- A PhD in Computer Science, Machine Learning, Mathematics, Statistics, or a related technical field is strongly preferred. We will also consider candidates with an MS and an exceptional research record.
- A demonstrated record of peer-reviewed publications in relevant conferences or journals, with substantive contributions to the research.
- Independently lead research, design rigorous experiments, and use results including negative or unexpected findings to guide next steps.
- Clear communication, including the ability to explain experimental design, findings, and limitations to technical and non-technical audiences.
NICE TO HAVES
- Research on deepfakes, presentation attacks or adversarial robustness.
- Experience with noisy, imbalanced real-world data.
- Experience shipping ML research to production.