Nalini Ramanathan
I am a first-year PhD student in Computer Science at NYU, co-advised by Erdem Varol and Pavel Izmailov. I work on neuro-inspired AI: using principles of neural computation to build machine learning models that are more efficient and robust. I am currently working on connectome-constrained modeling of the fruit fly in a robotics setting. Previously, I worked on hybrid spiking–nonspiking neural networks and connectome-constrained models of the fruit fly olfactory circuit.
I completed my MS in Computer Science at NYU in 2026. Before that, I spent five years at Altman Solon, first as a strategy consultant and then as a data scientist on the machine learning team, where I built computer vision models for utility pole detection and a Markov-model-based marketing attribution system. I did my undergrad in Economics at Dartmouth College, where I wrote my honors thesis with Treb Allen on how US services exports affect labor outcomes.
News
- Sep 2026Started my PhD in Computer Science at NYU, co-advised by Erdem Varol and Pavel Izmailov.
- Aug 2026Presented our extended abstract on hybrid spiking-nonspiking connectome-constrained models at CCN 2026.
- May 2026Received the Prof. Stuart Steele M.S. Student Award, given to one graduating NYU MS CS student per year for research and/or leadership.
- Apr 2026Defended my MS thesis at NYU, Biologically Constrained Heterogeneity: Hybrid Spiking-Nonspiking Connectome-Constrained Models.
- Nov 2025Presented work on directly connected hybrid spiking-nonspiking networks at the SNUFA Workshop 2025.
- Sep 2025Lfp2vec, on self-supervised localization of microelectrode arrays from raw LFP, was accepted at NeurIPS 2025.
- Sep 2024Left Altman Solon and started my MS in Computer Science at NYU.
Publications
Connectome-constrained models of the fruit fly olfactory circuit that mix spiking and nonspiking neurons, reflecting the heterogeneity of real neural circuits.
Lfp2vec: an audio-pretrained transformer, further self-supervised on unlabeled LFP recordings, that predicts the brain region of each electrode and generalizes across labs and probe geometries.
Hybrid networks that connect spiking and nonspiking neurons directly, neuron to neuron, and how well they generalize across tasks.
Earlier Work
- Labor Effects of Offshorable Service Exports in the United States
Nalini Ramanathan, Treb Allen. Carroll Round Proceedings XVIII, 2019. Undergraduate honors thesis, Dartmouth College. - The Role of Extracellular Matrix Molecules in Axon Regeneration
Nalini Ramanathan, Y. Yip. Pittsburgh Regional Science and Engineering Fair, 2013.