Cyril Marzook

Cyril Marzook

Research Engineer and Data scientist

Projects

I have been part of multiple projects related to Software Development and Artificial Intelligence. In this page you can find some of them. In case you want to know more about any of them, feel free to ping me via mail or linkedin.

LLM-Resume-Evaluation (AI job application process)

This repository contains personal implementations and it tailors resumes to match the unique requirements of specific job offers. This method ensures that each resume is optimized to stand out and resonate with the expectations of prospective employers.

Jigsaw Google : Toxic Comment Classification

This project is our participation to the Toxic Comment Classification Challenge, a Kaggle competition proposed by Jigsaw, a Google’s subsidaries belonging to Alphabet. In this competition, the challenge was to build a multi-headed model that is capable of detecting different types of toxicity like threats, obscenity, insults, and identity-based hate better than Perspective’s models in 2018. Here you can find the code and my report.

Deep Reinforcement Learning for finance

This projectimplements a Deep Reinforcement Learning (DRL) approach in the context of financial trading. The goal is to train an agent to make trading decisions based on historical financial data. The implementation leverages deep learning, specifically Long Short-Term Memory (LSTM) networks, and the Deep Deterministic Policy Gradient (DDPG) algorithm.

Optical Character Recognition using AI

This repositoryimplements an Optical Character Recognition algorithm that uses CNNs. It also contains a web app to draw some numbers and check the model accuracy.

Time Serie Analysis on Cooling days

I have developed this project in order to learn R and see what are the possible applications.

Kaggle Competitions

During my research master at PSL I participated in a Kaggle competition. Here you can find some information:

Ai for Alpha x PSL University : DRL for Finance

Decoding a financial strategy. This strategy is the aggregated performance of hedge funds performances. To decode this strategy, an agent is asked to provide each day the allocation (a percentage) invested in 11 factors. These factors are: