pabannier Goto Github PK

followers: 76.0 following: 11.0 repos: 60.0 gists: 8.0

Name: PAB

Type: User

Company: Owkin

Bio: Data scientist at Owkin | Mainly code in Python, C and C++

Twitter: el_pa_b

Location: Paris

Blog: https://pabannier.github.io

About me

I work as a Data Scientist at AI biotech Owkin.
Previously, I interned at INRIA Parietal on solving neuroscience (M/EEG) inverse problems.
I graduated from Ecole Polytechnique and HEC Paris with a double major in data science and management.

In 2022, I co-created skglm, a fast sklearn-compatible solver for sparse generalized linear models. More recently, I've become interested in fast inference for large language models. I have implemented Bark.cpp, a port of SunoAI's Bark model in C/C++, as well as specialized models like BioGPT.cpp.

Cool open-source projects I contributed to

MNE-Python, a toolkit for exploring neurophysiological data in Python
Linfa, the leading crate for machine learning and data analysis in Rust
Benchopt, a benchmarking suite for optimization algorithms

Other projects I worked on

Encodec.cpp, Meta's neural codec model ported in C++
SparseGLM, a fast coordinate descent solver in Rust
Nanograd, a lightweight deep learning framework built around Numpy arrays
NarrateMate.ai, a Next.JS web app to practice language comprehension listening to YouTube videos

PAB's Projects

abud

A tiny pub-sub lib for streaming

algo_fairness_ethics

Group project for the Algo fairness class

anomaly_detection

Implementation in Python of the anomaly detection algorithm taught in the Stanford Machine Learning course on Coursera

aptos2019-competition

source files for aptos 2019 competition

astronvim

AstroNvim is an aesthetic and feature-rich neovim config that is extensible and easy to use with a great set of plugins

automatic_hp_selection_for_meg

Automatic hyperparameter selection for Lasso-like models solving the M/EEG source localization problem

bark.cpp

Suno AI's Bark model in C/C++ for fast text-to-speech

benchmark_elastic_net

Benchmark repository for the elastic net problem

benchmark_lasso

Benchopt benchmark for Lasso

benchmark_lasso_path

benchmark_logreg_l1

Benchopt benchmark for Sparse Logistic Regression

benchmark_multi_task_lasso

Benchopt benchmark for Multi-Task Lasso

benchmark_multitask_lasso

benchopt

Making your benchmark of optimization algorithms simple and open

bengali-ai-competition

kaggle bengali ai competiton source files

biogpt.cpp

Port of Microsoft's BioGPT in C/C++ using ggml

coding-exercises

My implementation of useful data structures, algorithms, as well as my solutions to programming puzzles.

cs_notes_tips_tricks

A bunch of notes about CS, ML, networking that I usually learn on the job.

deep-learning

🦅 Deep Learning awesome cheatsheet

dummy_benchmarking_rust

a dummy repo to learn how to effectively benchmark in rust

encodec.cpp

Port of Meta's Encodec in C/C++

fast-melanoma-detection

A malignant melanoma classifier with deep learning deployed in a Streamlit web app

fgfr3mut

Official Implementation of the "AI allows pre-screening of FGFR3 mutational status using routine histology slides of muscle-invasive bladder cancer" paper