SANKET KACHOLE

I am a Postdoctoral Research Fellow at University College London, working on multimodal artificial intelligence for early detection of oral cancer. My research focuses on integrating genetic data, clinical information, and medical imaging to build models that can support clinicians in detecting disease earlier and more reliably. I am particularly interested in how we can design AI systems that are not only accurate, but also robust, interpretable, and clinically useful.

My work sits at the intersection of machine learning, computer vision, and healthcare. I have experience across the full research pipeline: curating and annotating datasets, designing model architectures, training and evaluating models, and working closely with domain experts to translate technical results into practical impact. I have worked with a range of methods including generative models, transformers, multimodal networks, graph neural networks, and spiking neural networks.

I completed my PhD in Artificial Intelligence at Kingston University, where I focused on neuromorphic vision and event-based perception for robotics. During my PhD, I developed algorithms for event-based segmentation and panoptic perception, combining asynchronous event streams with conventional RGB video. This work led to several publications in journals and top-tier computer vision venues, including a Best Paper award at a CVPR workshop.

Alongside my academic work, I have industry experience as a Data Scientist at homeprotect (home insurance) and Santander UK. At homeprotect, I built computer vision and NLP systems for automating claims assessment, using transformer-based models for both image segmentation and text classification. At Santander, I developed machine learning models and data pipelines for financial crime risk, focusing on robust, production-grade code and model monitoring.

I enjoy working in multidisciplinary teams with clinicians, engineers, and researchers to tackle real-world problems. My broader interests include event-based and neuromorphic sensing, multimodal learning, representation learning for healthcare, and AI systems that can move from controlled lab settings into noisy, real-world environments. I am always open to collaborations at the interface of AI, medicine, and robotics.