TL;DR

I am a fifth (and final) year PhD student in Bioengineering at the University of Pennsylvania, advised by Dr. Flavia Vitale, in the Center for Neuroengineering and Therapeutics. During my PhD, I have obtained extensive experience working with big datasets and developing machine/deep learning frameworks with applications in computational neuroscience and brain computer interfaces. My work has been published at top-tier AI conferences (NeurIPS) and journals (Science Translational Medicine).

Looking for full-time opportunities

I enjoy tackling multidisciplinary challenges and I am looking for full-time opportunities in the AI/ML space starting September 2025.

Current Projects

Multi-subject neural decoding from sEEG. I am currently developing methods to enable cross subject neural decoding from stereotactic EEG data. Currently, neural decoders based on sEEG are designed to be subject-specific, due to the challenge of integrating inter-subject data, where each subject gets implanted with a variable number of electrodes placed at distinct locations in their brain. However, single-subject models don't generalize to new subjects and can't scale due to within subjects data limitations. Leveraging the versatility of transformers, I am constructing unified models for sEEG decoding that can be trained across subjects, despite the number and placement of electrodes in each subject's brain. This approach enables me to combine datasets collected across multiple subjects and medical centers, to train more expressive models on orders of magnitude more data. The goal of this work is to enable robust, high-performance neural decoding from sEEG in a scalable way. [Project Page]

Energy efficient neural decoding from microelectrode arrays. BCIs have revolutionized the quality of life of people suffering from neurological disorders by enabling them to control computers with their thoughts, ultimately restoring their digital freedom. While extremely helpful, most BCIs rely on artificial neural networks (ANNs) that are power hungry, requiring users to recharge their BCI every few hours, which is impractical and frustrating. To combat this, I am developing neural decoders based on spiking neural networks, which are very energy efficient compared to ANNs when deployed on neuromorphic hardware. The goal of this work is to enable high-performance, energy efficient neural decoding from microelectrode arrays, to enable day long BCI use without the frustrating need for frequent recharging.