I'm an Electronic and Information Engineering student at Imperial College London, with a passion for developing AI applications to increase efficiency of processes.

I come from France, but have lived in Vienna, Austria and London, United Kingdom.

I speak French, English, Spanish and German.

I love travelling and have been to 40 countries/territories so far!

• Software Development Engineering Intern at Amazon

  • Built a full-stack modern web application
  • Integrated a React frontend with an API Gateway + Lambdas backend
  • Built a full end-to-end user authentication pipeline
  • Used CI/CD development techniques

• Software Engineering and Data Science Intern at Equinor

  • Used Natural Language Processing (NLP) to automatically find a relevant document betweens thousands
  • Implemented a custom Selenium web scraper to dynamically load webpages and extract data to form a document database
  • Using Retrieval Augmented Generation (RAG), leveraged an LLM to generate an answer to a question based on relevant documents extracted from the database using NLP
  • Hosted the solution in the cloud using Docker and Kubernetes
  • Created a ChatBot type application using HTML/CSS/JavaScript and Python for an easy interface

• FPGA Accelerator for parallel computing

  • Implementation of Conway's Game of Life on a PYNQ-Z1 FPGA board using custom hardware
  • User Interface using Computer Vision to detect drawing movements and pause signals
  • Real time evolution of a 1280x720 grid at a maximum of 200 000 evolutions per second

• Music Synthesiser with DOOM

  • Music synthesiser capable of playing notes, with volume and octave control, recording functionality, changing waveform abilities
  • Graphical user interface to interact with the various functionality
  • Fully working DOOM-like 3D game with purpose-built game engine, obstacles, enemies, shooting functionality, and chunk-based world generation
  • All running at a maximum CPU utilisation of 48%, DOOM running at 50fps (20ms refresh rate)

• PiTrainer: AI-Powered Personal Gym Coach Embedded System

  • Android and iOS compatible mobile app in React Native
  • AWS EC2 instance backend server in Python Flask to handle HTTP requests and Dynamo database storage
  • Custom-trained machine learning model to analyse gym rep quality
  • RaspberryPi in Python to collect accelerometer and magnetometer data, analyse reps using signal processing methods

• 32-Bit RISC-V ISA CPU Design and Implementation

  • Fully functional CPU with pipelining, data cache and automatic hazard handling and branch prediction
  • Use of SystemVerilog, C++ and RISC-V ISA
  • Comprised of a Program Counter, Control Unit, Hazard Handling Unit, Data Memory RAM, Instruction Memory ROM, Data Cache and Pipeline Registers

• 2 Player Flight Simulator with wireless FPGA controller

  • Features 2 DE10-Lite FPGA controllers, using switches, buttons and onboard accelerometer to control a plane
  • FPGA controllers wirelessly connect to AWS EC2 instance to asynchronously process inputs and output next plane state
  • State of both planes rendered in the same 'world' with Unreal Engine 4 to provide smooth gameplay
  • The goal of the game is to use the FPGA as a joystick to control and race the other plane through a portal
  • Time taken is recorded and stored as a highscore in a DynamoDB AWS database

• Remote Controlled Rover

  • Functioning rover robot capable of movement in 4 directions
  • Remote controller with Wifi communication
  • Capable of measuring magnetic fields, decode radio signals and decode infrared signals
  • Dual Arduino board with I2C protocol for multi-threaded control, sensing and filtering
  • 3D Printed frame

• Autonomous search-and-rescue quadcopter

  • 4-rotor Drone capable of detecting and avoiding obstacles autonomously
  • Use of the YOLOv3-tiny convolutional neural network to automatically locate obstacles in field of view
  • Custom made algorithm using simple trigonometric principles to find the shortest path around the obstacle
  • Designed to travel from a Point A to a Point B as fast as possible in Alpine search-and-rescue operations

• RISC-V ISA C-Language Compiler

  • Implemented in 36 hours
  • Implemented functions and function calls, variables, data types, arithmetic and boolean operations
  • Capable of compiling a C-90 program into RISC-V Assembly