October 2017 – March 2019: Airbus Defense and Space (Meritis PACA subcontractor)
My first real work experience was at Airbus Defense and Space Sophia Antipolis, as a subcontractor for Meritis PACA. There, I worked on two projects:
- Building an automatic aerial imaging processing pipeline for drone imagery on Google Cloud Platform, using C++, Bash, Go, and Python.
- Building a semantic segmentation pipeline for 3D models using Deep Learning
3rd year intership: TIMC-IMAG Laboratory (March-September 2017)
This internship was my Master’s Thesis, entitled “Hip fracture surgery simulations and patient-specific muscle mesh generator”, under the supervision of Matthieu Chabanas (TIMC-IMAG) and Medhi Boudissa (CHU Grenoble).
We created a patient-specific surgery simulator based on CT scans and a biomechanical model of the hip, using Artisynth, a physics simulation framework in Java.
Then, we worked on a muscle mesh generator using finite elements and included it in the simulator.
Report and slides are available here.
2nd year internship: ARM Cambridge (Summer 2016)
During this internship, I worked on low level Computer Vision and Image Processing. I contributed to the ARM Compute Library (available here), an image processing and machine learning library designed for ARM processors. This library can be up to 15 times faster than competing ones, such as OpenCV.
I implemented and optimized image processing algorithms in C/C++ using ARM NEON (Advanced SIMD), created an Android app to demonstrate the capabilities of the library using Java, C++ and JNI, and created a dice recognition demo using OpenCV and Support Vector Machines.
I also took part in the ARM Global Intern Challenge, which involved developing for the ARM Cortex-M platform (BBC micro:bit).
1st year internship: CEA Grenoble/BIG/Biomics (Summer 2015)
This internship was my first experience in a research lab. Under the supervision of Laurent Guyon, I created and optimized an R/Bioconductor package for the analysis of High Throughput Screens. Significant time gains were obtained on the in-house pipeline the team is using (from several hours down to a few minutes).
My internship report is here.