Fabien Amazo

- Membre de l’équipe qualité, en charge de définir et tester les différents produits STET avant leur passage en production. - Utilisation de Bash pour l’automatisation des tests, PostgreSQL pour la gestion des bases de données, et Python pour le développement de scripts de test.

- Gestion des tickets techniques liés aux transactions bancaires, analyse des logs et résolution des problèmes. - Collaboration avec les équipes techniques pour la mise en place de solutions logicielles. - Rédaction de documentation technique et support aux équipes de développement.

As a member of the R&D team, I have been in charge of designing and implementing the control methods for the different actuators on automated forklift trucks. Tasks : - Weekly meetings to set the goals. - Presenting and designing the solutions. - Debugging and troubleshooting the SDK and existing products by identificating, reproducing and solving the problems encountered by the fields engineers. - Testing solutions with adapted scenarios on available trucks. - Solutions submitted through git for validation. - On site debugging when necessary (Boston, Tracy, ...).

Spatio-temporal facial expression recognition from video sequences based on deep learning techniques. Facial expressions are muscles movement and configurations which in addition to body language, voice tone and other non verbal signals will reflect the emotional and mental state of an individual. This project's main goal was to explore and compare several dynamic methods (taking the temporal dimension into account) to estimate emotions based on deep learning. Withing the laboratory, constant communication with its members, the understanding and handling each item of the processing chain were some of the elements of the daily environment. Tasks : - Conducting a technological survey on state of the art methods for deep learning based facial expression recognition (papers analysis, deep learning specialized forums, ...). - Researching and analysing the different available databases. - Designing a generic code to preprocess the data. The goal here is to extract and align faces as well as a normalization in order to have the best possible data quality to feed our neural networks. - Designing the different networks to compare. - Separating the data into different sets (Training, test, validation). - Applying the different chosen methods. In this phase, the main challenge is to generate the training algorithms, as we have to deal with videos, we need to optimize the use of memory (RAM for loading image batches) as well as well dimension our model (GPU to load our model, to not use all the shared space). Python / Tensorflow / Keras / Scikit-learn / OpenCV / CNN, RNN, RCNN

First year, conception and implementation of the data infrared emitting-receiving module : - Choice of the technical solution.- Design and dimensioning of the electronic board and components. - PCB design and integration of the solution into the robot. - Participating in the robot assembly. Second year, design of the hardware and software part of the robot controls : - Choice of the technical solution.- Design and dimensioning of the motor board and its electrical components. - PCB design and integration of the robot solution. - Programming of the robot positioning equations (first the wheels and the complete robot), then coding and testing of the different movement strategies. C, Altium Designer (PCB design), Code Warrior (embedded computing)

Design of a module for the replication of frames to control nano-materials drops shape from a nano-materials printer. The Lab possessing a nano-material printer, one of the objectives of the project consisted in exploring different methods of expulsion loaded drops. The goal of this project was to generate an electric signal allowing to control the volume of the room of expulsion of the drops. Tasks : - Analyzing the model of the frames to reproduce. - Choosing the solution, an Mbed micro-controller with an amplifier. - Coding the solution on the board and integrating the amplification module. C++, Mbed