Framework for training and evaluating self-supervised learning methods for speaker verification.

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

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

Ray Tracer written in C++/CUDA enabling real-time photorealistic rendering of multiple shapes with lights, shadows, and mirrors.

A deep neural networks framework similar to Keras and written from scratch in C++/CUDA.

Machine learning pipeline to classify hyperspectral images of fields.

Concurrent dictionary with one implementation based on concurrent hash maps and one based on concurrent tries.

Reprogramming of the memory C allocation tool (malloc, realloc, free)

Python bindings for Pylene, a C++ image processing library.

Deep learning methods to classify images of ships.

Chess engine and AI written in C++ to compete with other students.

Fully functional shell in C. Variables, paths, globbing, jobs... handled. Interpreting done with an LL parser and an AST