I am an artificial intelligence engineer, graduated from the University of Sherbrooke in Canada and from the Telecom Nancy engineering school in France.
I'm also a former president of the Junior-Entreprise Telecom Nancy Services, which has about thirty members and provides IT services.
I studied in four universities, and learned many different programming languages and working methods that have made me a versatile engineer able to temporarily handle web programming (front and back), or software development tasks.
I'm looking for a job in which I can deepen my expertise in AI, whether it's through the creation of machine learning models (prediction, classification, time-series, etc.), or data analysis.
2021 - 2022
Three semesters leading to graduation. The description of the subjects studied is given below:
Neural networks: SVM, bagging, boosting, classical neural networks, convolutional networks, deep learning models (VGG, ResNet, UNet), recurrent neural networks, generative models
Descriptive data analysis: clustering, association rule extraction, statistical description, similarity measures, anomaly detection
Math for AI: Bayesian and Markovian networks, maximum a posteriori inference and simulated annealing, Monte Carlo methods, time series and latent variable models.
Database modelling: Distributed database design. Data warehousing and data mining. Translation of an object-oriented model into a relational model.
2019 - 2022
Data visualization, and data analysis (principal component analysis, linear discriminant analysis...) with R and Python
Data mining, indexing with Lucene, Spark, Hadoop, MongoDB, Elasticsearch
Database modelling, MySQL
Parallel algorithms with MPI and OpenMP
Search algorithms (MCTS, MinMax)
Project management, management
Probability model, statistics, optimization.
Systems, networks and image processing.
2019 - 2019
One-term school exchange in computer science.
Algorithm courses: Complexity, dynamic algorithms (Floyd, chain matrix multiplication...), divide and conquer algorithms, Las Vegas and Monte Carlo algorithms, probabilistic algorithms (Sherwood, numerical algo), etc.
Advanced web technology: TCP, UDP, Sockets, Addressing and routing, Laravel, Symfony
Data structure and C++
Video game design course
2017 - 2019
Two years of study in which I acquired the necessary knowledge to become a developer.
Web, PHP, Laravel, Javascript
OOP, Java, C#, Design Patterns
System, Memory Management, segmentation, thread
Project Management
Database Management, SQL
Download report cardMay 2022 - August 2022
Carried out various deep learning tasks in a start-up specialised in AI, with a team of about ten experts.
Creation of models such as classical neural networks, convolutional networks, transformers, or auto-encoders with Pytorch and TensorFlow. For each project, pre-processing of the data was performed.
Use of various data types (time-series, images, audio, textual data, etc.)
Production of visualisations for clients.
But also redesign of the website (front-end) with VueJS (https://explor.ai), creation of the back-end of the compagny's management tool, creation of the corresponding database architecture, and API testing with Postman.
Mai 2021 - Août 2021
Database migration from Oracle to Postgresql.
Testing for more than a month and a half on the new database.
Updated C# projects to verify proper data access and management at the software level.
Septembre 2020 - Mars 2021
In charge of the management of the Junior-Création Telecom Nancy Services, which has about thirty members and provides IT services.
Ensures the coordination and the good functioning of all the departments (HR, treasury, project managers, secretariat...) as well as the sustainability of the Junior and the development of the relations with the external actors.
Responsible for the proper drafting of the strategy, the action plan and ensures its implementation.
February 2020 -
In charge of the management of the internal divisions (HR, audit, communication, IS) of Junior-Création Télécom Nancy Services.
November 2019 - February 2020
Responsible for customer retention, business development, and the definition and follow-up of the commercial strategy.
2019
2018
Test obtained with the score of 875/990.
2016
2016
Basic first aid training.
2013
This diploma raises awareness of the dangers of addiction, and accredits the title of referent student to help people who are prone to addiction.
Click on the project to learn more !
Inferring a person's emotional state from textual data using two methods: transformers (Hugging Face) and a combination of embedding with Word2Vec.
Inferring a person's emotional state from text data using two methods: transformers (Hugging Face) and a combination of embedding with Word2Vec.
Creating a speech recognition system with a simple neural network to help people with visual or physical disabilities select the right lift control.
Creating a speech recognition system with a simple neural network to help people with visual or physical disabilities select the right lift control.
Testing the use of the AWS Transcribe API for speech recognition.
Kaggle whale identification challenge with convolutional networks. Bounding box with YOLO to find where animals are located on the image. Image processing, data augmentation, EffNet model with Arcface
The objective of this Kaggle challenge is to automate the photo-identification of whales and dolphins. These animals can be uniquely identified by their dorsal fin.
The data from this competition contains images of over 15,000 unique marine mammals from 30 different species.
The pictures of these animals are of different resolution and brightness. In addition, some images are taken from a distance, with several animals, or with other elements on top such as a boat or people.
We then used YOLO's Bounding box to detect the animals and then crop the images. Finally we had to resize all the images.
We then augmented the data by rotating, translating and shearing the image to mimic photos taken from different angles. We also augmented the data by adding noise, or changing the brightness and contrast. Regarding the models implemented, we tried a 152-layer ResNet, and a b0 EffNet.
We also tried to use Arcface which allows to give the similarity measure of 2 faces together and to apply it to our problem.
Using the Encoder-Decoder, UNet, and MSCADC models to perform segmentation on the Automated Cardiac Diagnosis Challenge .
Using the Encoder-Decoder, UNet, and MSCADC models to perform segmentation on the Automated Cardiac Diagnosis Challenge .
Creation of an enterprise calendar interface with VueJS.
Creation of an enterprise calendar interface with VueJS.
Data reduction, learning with a convolutional neural network, study of data distribution in embedding.
Data reduction, learning with a convolutional neural network, study of data distribution in embedding.
Pre-processing of data. Analysis, selection and visualization of the correlation of the best attributes. Principal component analysis.
Attribute analysis, pre-processing of data, Selection of the top 10 attributes, and proposing a visualization of them. Evaluation of correlation relationships between the top 10 attributes. Determination of the optimal number of principal components via PCA.
Create a relational database for a restaurant and then establish a data warehouse for customer order queries.
Create a relational database for a restaurant and then establish a data warehouse for customer order queries. Finally, create a temporal database to historicize the data.
Using classical data analysis and AI methods to format and classify tree leaf images in Python.
Using classical data analysis and AI methods to format and classify tree leaf images in Python. Preprocessing, Neural Networks, Support Vector Machine, K nearest neighbors. Implement Boosting (AdaBoost), and Bagging (RandomForest, combination of classifiers) algorithms.
Extract association rules on a banking database in Python.
Extract association rules on a banking database in Python to determine, from data such as salary, age, being married, etc., the characteristics of customers who have a People's Savings Plan (PSP) and the customers who could benefit from a PSP.
Implemented two AIs in Python using Alpha-Beta pruning algorithms and Monte Carlo Tree Search (MCTS).
Implemented two AIs in Python using Alpha-Beta pruning algorithms and Monte Carlo Tree Search (MCTS) to compete against each other and against players in the game of Chinese checkers.
Creation of a java compiler for an object language created for this project. Creation and manipulation of a grammar to allow the language to be translated into machine language and then used.
Creation of a java compiler for an object language created for this project. Creation and manipulation of a grammar to allow the language to be translated into machine language and then used.
Research project for the CRAN Laboratory (affiliated to the CNRS) aimed at using mixed reality to help operators in their product regeneration tasks. We used Unity to create an interface for Hololens glasses.
Research project for the CRAN Laboratory (affiliated to the CNRS) aimed at using mixed reality to help operators in their product regeneration tasks. We used Unity to create an interface for Hololens glasses.
Integration of several sources of heterogeneous and very large biomedical data into a system that allows to know, from a symptom, the diseases causing this symptom, the drugs to be used to cure it and the drugs causing this symptom.
Integration of several sources of heterogeneous and very large biomedical data into a system that allows to know, from a symptom, the diseases causing this symptom, the drugs to be used to cure it and the drugs causing this symptom.
Creation of a global data schema to map the relationships between data sources.
Creation of an application, using Javafx, to achieve the desired functionality.
Game made with Unity in which 2 players compete by asking multiple choice questions.
Game made with Unity in which 2 players compete by asking multiple choice questions. The bad answers inflict damages. A connection menu to servers was also created for this game.
Realization of a hiking application in JavaFx during a coding week.
This software was made in Java using mapjfx during a coding week.
The application offers a number of features for organising hikes.
In particular, it allows you to visualise walks (on a map or in a list) and to have information on them such as their length, difficulty or altitude difference.
It is also possible to save these projects in PDF. A summary of the route and the various checkpoints, accompanied by a map, is then saved on the device.
The application allows you to create routes by entering destinations, importing a GPX file or drawing on the map.
Finally, it is possible to save a route in your favourites, change the topology of the maps and other parameters.
Software, made in C, recognizing whether a film commentary is positive or not using machine learning.
This software made in C uses a database of comments as found on Rotten Tomatoes to learn whether a given comment is positive or not. We have also proposed a comparative performance study between hash table and radix trie for this particular use. This software has been realized as part of the module "Projet Pluridisciplinaire d'Informatique Intégrative" at Télécom Nancy. For this project we put agile software development into practice by working by sprint, among other things.
Game created with Unity as part of a 24-hour game jam organized by Telegame Design.
This game was created with Unity as part of a 24-hour game jam organized by Telegame Design. The theme was "The more you have, the worse it is". Our basic idea was to make a game with a factory that produces more and more cows. The goal of the game is to make sure that the cows don't fall into the surrounding cliffs, but obviously the more cows the worse it is. According to the rules of the game jam, all textures, sounds, etc. had to be made by ourselves. All the textures were made in Photoshop and the materials in Blender.
Implementation of several steganographic methods (LSB, BPCS, etc.), evaluation metrics (EQM, SSIM, etc.), and realization of a research report on the subject.
This project, carried out according to agile software development methods, allowed us to put into practice different steganographic methods (LSB, BPCS, etc.), and evaluation metrics (EQM, SSIM, etc.). But we were also able to draw up a research report of about fifty pages on the subject, allowing us to acquire a certain mastery in this field.
Remake of the video game "Smash Bros Ultimate". Realised during a coding night, continued thereafter. Made with Unity.
Smash Reborn is a game in which 2 players compete against each other. Each player tries to push the other out of the platform with an attack that ejects the other player. This game has been made during a coding night, i.e. during a dozen of hours, in a team of 4. This is my first game made with Unity.
2D video-game, made in C++ with the SFML Library.
Infected prison is a 2D survival game. The player sees the game from a High-angle shot (i.e. directly above the action). Players will have to survive in a zombie apocalypse context. To do so, they will have to collect objects and materials to defend themselves. This game is therefore aimed at fans of retrogaming, or 2D games. Amateurs of survival and crafting games will also be interested. The game can be played alone or in multiplayer in cooperation.
Carpool web application.
Travel Express is a web application that allows you to quickly find a carpool near your home. It also allows you to manage your own trips as a driver, or to plan trips as a passenger. Thanks to Travel Express, the user can define his or her preferences, both as a driver and a passenger, and is able to obtain all the information necessary to organize a carpool in a few clicks.
Encryption software.
Application made in C allowing to encrypt, decrypt and craking messages. The generation of primes, the Miller-Rabin test, the RSA and Rabin-karp algorithms, etc. are among the concepts seen during this project.
Analysis and statistics report and creation of patient files based on data from hospitalized persons. Realized with SAS.
Simulation of the management of a garden shared by an association and all the challenges it represents.
Project carried out under the supervision of Victoria Saché, expert in sociology more particularly in the context of gardens. Simulation of the management of a garden shared by an association and all the challenges it represents (management of the members of the association, financial management, promotion of the garden at the town hall etc...).
Language learning app inspired by Babbel.
Language learning app inspired by Babbel.