Eklavya Sarkar

Eklavya Sarkar is a PhD candidate in Machine Learning at EPFL and Research Assistant at Idiap Research Institute, working in the Speech and Audio Processing group under Dr. Mathew Magimai Doss. His research focuses on speech and audio methods for analyzing human and non-human vocal communication, particularly in the context of the Swiss NCCR Evolving Languages project. He holds a MSc in Data Science from the University of Bath (First Class) and a BSc in Computer Science from the University of Liverpool (First Class).

Education: PhD in Machine Learning (EPFL, 2021–present), MSc Data Science (University of Bath, 2018–19), BSc Computer Science (University of Liverpool, 2015–18). Key research areas include self-supervised learning, bioacoustics, generative adversarial networks, and biometric security. Previously interned at Idiap (biometrics), CERN (CMS Experiment), and conducted exoplanet research with Nobel laureate Didier Queloz.

Research interests span cross-domain transfer learning for bioacoustics, speaker identification in animal vocalizations, and vulnerability analysis of face recognition systems. His work bridges machine learning theory with applications in wildlife communication analysis and cybersecurity. Recent publications emphasize self-supervised learning models for marmoset vocalization analysis and face morphing attack detection using GANs.

Awards include 3rd Prize in the International Create Challenge 2020 for adversarial attack detection. Organized the Perspectives on AI Symposium series and led student housing committees. Technical expertise spans Python (PyTorch, Hydr), signal processing, and multiple machine learning frameworks. Engaged in interdisciplinary projects at the intersection of physics, computer science, and linguistics.