Nossa Claim this profile

Episodic memory for AI agents. Records your screen locally, compiles it into structured activity frames, serves them over MCP. No cloud, no LLM in the loop.

Risk on asset management, prediction for up and down movements using Deep Neural Network

Some of the work achieved during the Mistral AI hackathon. Our platform provides a solution based on Mistral LLM to facilitate easy access to every existing patent related to your field using RAG.

Creation of shallow and deep neural network from scratch. Model used for the quantitative trading project (cf nossa-y/quantitative_trading/etude2_DNN1/predictions.ipynb)

Solution for sales prediction. (including description of the project)

Solution for a supply chain company. (including presentation)

HOUSTON - mission control for computer-use agents. Never trust, always verify. Built at The Computer Use Hackathon (H Company, SF, July 2026)

My personal deposit

Exploring and illustrating the theory of martingales, brownian motion, Heston model, Black Scholes model, Markov chain through simulation and analysis

Nocta releases - get it at usenocta.app

No description.

No description.

No description.

No description.

No description.

📲 An agent for sourcing, curating, and scheduling social media posts with human-in-the-loop.

No description.

Yolo based app finetuned on orangerie museum portraits, implemented with executorch

Testing different Embedding Models, from Architecture to Implementation

Here is our solution to build an innovative collaboration between hospitals and Research Organization. (1st prize winner at HuggingFace x Zama hackathon)

ML algorithms designed to operate directly on fully homomorphic encrypted data.

Here is the implementation of an end-to-end Machine Learning project with Docker and CI/CD

Unsupervised learning for objects clustering

Study on different supervised learning on real world datasets

No description.

Here are articles I co-authored and authored to complete student projects

A code repository designed to show the best GitHub has to offer.

Some ML work I carried out during my internship at Odecloud, San Francisco, California in summer 2023

Local implementation of a Mistral LLM (version Q4_K_M). In this repository, I provide a tutorial on how to implement a Large Langage Model locally and 2 ways to use it.

To learn how to use fastapi which is the main tool used by my last internship developers, I built some alpha apps with it.

This repository contains codes of data processing, optimisation function and model implementation for predicting wait time for transport. (including presentation)

Noisy trajectory prediction and filtering using the Kalman filter. (presentation included)

Swap web app. Python code is the optimization function enabling users to exchange their money while minimizing transaction fees.

The project "behave&earn" aims at making a content producer rewarded for its accomplishment as soon as he completed a job, through smart contract deployment and oracles. (including presentation)

Bias-free model for candidate recommandation. (including presentation)

Some of my personal projects