Speaker diarization with identity and language insights, via browser or API, powered by SOTA AI speech models.

Toolkit for training and evaluating Self-Supervised Learning (SSL) frameworks for Speaker Verification (SV).

SOTA method for self-supervised speaker verification leveraging a large-scale pretrained ASR model.

My completely customizable personal website built with Gatsby and sourced from static markdown files.

Generate Japanese estampes from handmade sketches based on Pix2Pix model architecture.

Implementation of SSL frameworks for Speaker & Language Recognition in Tensorflow / Keras.

Tiny automatic differentiation (autodiff) engine for NumPy tensors implemented in Python.

Ray-tracer, written in C++, featuring different objects, lightings, textures and scene configurations.

A tiny neural network framework developed entirely from scratch in C++ and CUDA.

3D multiplayer game made with Unity where one player drives while the other is building the road.

Optical Character Recognition software based on a simple NN and written from scratch in C.