Projects

These are some of the projects I have taken up during my career.

Swiss-Knife for electrophysiology

Swiss-Knife for electrophysiology

Built a production-style data pipeline for large-scale time-series analysis — from raw signal ingestion through feature extraction, dimensionality reduction, and unsupervised clustering. Processed 3.6+ billion data points across 1200+ recordings. Implemented with reproducibility-first principles including version control, containerization, and FAIR data standards. Delivered as part of two peer-reviewed publications.

Learn more →
Heterogeneity in Reservoir Systems

Heterogeneity in Reservoir Systems

Collaborated with Ghent University to design and evaluate neural network architectures with heterogeneous temporal dynamics. Benchmarked memory capacity, stability, and dimensionality across standard tasks, showing that architectural diversity improves performance. Built and trained models in TensorFlow 2.x on HPC infrastructure.

Learn more →
Reverse Engineering Recurrent Neural Networks

Reverse Engineering Recurrent Neural Networks

Different neural network architectures solve the same task in fundamentally different ways. I systematically trained 100+ recurrent and spiking neural networks on a memory task and reverse engineered their internal dynamics using fixed-point analysis and dimensionality reduction to understand how design choices shape learned representations. Run on GPU-enabled HPC infrastructure using CUDA and SLURM.

Learn more →
GLIF Model Fitting

GLIF Model Fitting

Migrated a neuron model fitting codebase from Python 2 to Python 3 and extended it with a BluePyOpt-based optimization variant, adding evolutionary algorithm support alongside the existing approach. Applied the refactored toolbox to fit biophysical models to experimental electrophysiology data, closing the loop between model development and real data validation.

Learn more →
Siamese Neural Network for extracting neural signatures

Siamese Neural Network for extracting neural signatures

I implemented a Siamese neural network to extract meaningful signatures of neurons recorded across two different patch clamp protocols.

Learn more →
Autonomous Warehousing

Autonomous Warehousing

In this project, we developed an autonomous warehousing system that utilizes advanced robotics and AI algorithms to optimize inventory management and streamline the supply chain process. I was responsible for designing the path planning algorithms for the autonomous robots, ensuring efficient path planning and obstacle avoidance.

Learn more →
Fitting Fritz-Hugh Nagumo Model on fMRI Data

Fitting Fritz-Hugh Nagumo Model on fMRI Data

In this project, I fitted the Fritz-Hugh Nagumo model to fMRI data. I used a parcellated fMRI dataset and ustilized FHN based network to understand connectivity.

Learn more →
Manifold Alignment for Neural Data

Manifold Alignment for Neural Data

I utlized UNICOMM algorithm to compare neuronal properties recorded across two input protocols by aligning the high-dimensional mainfolds.

Learn more →
Brain2 based neuron models

Brain2 based neuron models

This was a side project to implement common neuron models in order to fit them to noisy single cell electrophysiological data

Learn more →