AI Research Scientist (Postdoc)
CNRS (I3S Laboratory) (Jan ’26 - Present)

This work is a part of the ANR NAMED project, in collaboration with partners in neuromorphic hardware and vision.

Working on neuromorphic computer vision, with a focus on event-based vision and spiking neural networks.

• Studied the effect of human gaze-based filtering on action recognition in terms of accuracy and efficiency, using sparse 3D convolutional networks on RGB and event data (ICANN 2026).
• Worked on saliency prediction in event-based data, towards replacing human gaze with a learned predictor (ICONIP 2026).

Doctoral Researcher
Grenoble Institut des Neurosciences and INSERM (Mar ’21 - Dec ’24)

This work was a part of the ANR Brain-Net project, where I collaborated with both academic and industrial partners.

Developed a frugal Spiking Neural Network (SNN) for the unsupervised classification of multivariate temporal patterns.

• Applied the network to the classification of vocal and speech patterns, and to multichannel spike sorting of neural recordings.
• Published the work in Nature Communications, and filed a European patent application for the spike sorting algorithm.
• Presented the work at Neuroscience 2021, Neuroscience 2022 and CORTICO Days 2022.

Artificial Intelligence Research Internship
Grenoble Institut des Neurosciences and INSERM (May - Oct ’20)

This work was a part of the European Union Horizon 2020 Project.

Developed an unsupervised Spiking Neural Network (SNN) with Spike-Timing-Dependent Plasticity (STDP) for classification of vocalizations.

• Encoded raw analog audio into discrete spike trains with ‘time-to-first-spike’ encoding.
• Implemented a Low-Threshold-Spiking (LTS) Neuron model to mimic the activity of biological neurons by introducing a temporal dimension to the activation of neurons.
• Implemented the STDP learning rule to enhance learning by updating the synaptic weights of the network.

Research Internship
BrainTech laboratory and Université Grenoble Alpes (May - Oct ’18)

Analyzed vocalization data and reduced its dimensionality with Machine Learning algorithms like Principal Component Analysis (PCA) in order to facilitate mapping with cortical activity; critical for the development of a Brain-Computer Interface (BCI).

• Analyzed Minipig Vocalization Data (MVD) to facilitate mapping between MVD and cortical data.
• Studied spectrograms and labelled audio files of the vocalization data as grunts, squeals and screams using Spike2.
• Performed data cleaning and dimensionality reduction of the data through spectrogram analysis and noise removal.
• Coded functions to implement dimensionality reduction algorithms like PCA and t-SNE using MATLAB.
• Techniques Used: Spectrogram Analysis, Principal Component Analysis, t-SNE, Independent Component Analysis

Summer School

BrainCom Summer School - Barcelona, Spain. (September 2018)