Experience
|
10/2020 - Now: IP Manager
Gamestream, Ludres France.
- Developed an inference library for convolutional neural networks(CNN) in C++ with SDL 2.0 and CUDA.
Optimized CUDAs kernels using Nvidia NSight Compute to meet real-time performance constraints.
- Developed a framework in Python with Tensorflow and FFMPEG to create learning datasets and CNN from video
streams and XML specifications. It simplified the design and the comparison of differents CNN architectures.
- Developed CNN that solve real-time super-resolution for single images and video streams, improving the
compression rate of video streams.
- Protoyped a client/server for application based on OpenGL commands streaming in C++ improving performances
of 2D application on low-end hardware.
- Wrote proposals and reports about research done by a team of eight, approved at 100% for Research Tax Credit.
- Wrote tutorials and technical documentation for clients and new team members about the company’s stack.
02/2020 - Present: Freelance
Freelance, France
- Wrote feasibility study for non technical audience about 3D printing of a rigged characters in an arbitrary pose.
- Prototyped a pixel art drawing tool using C#, WPF and OpenGL. Developed an algorithm that computes normal
map from 2D drawn characters using example-based texture synthesis technique or CNN.
- Developed a web tool to previsualize a 3D printed model before printing it. I used WebGL to implement
well-known real-time computer graphics techniques: deferred shading, normal mapping, SSAO
03/2018 - 09/2019 Post-Doc
Université de Montréal, Canada. Supervised by Bernhard Thomaszewski
- Developed a path planning algorithm generating support and improving clay-based 3D printer in C++, reducing
the failure rates of ceramic printers (published at SIGGRAPH Asia 2019).
- Prototyped an application to measure the deformation of an object printed with a clay-based 3D printer from a
video timelapse using OpenCV.
- Experimented with cage based shape deformation (mean value coordinates and harmonic coordinates) in C++ and
with libigl for orthosis customization.
06/2017 - 03/2018 Software Engineer
Software Engineer, AddUp, Clermont Ferrand.
- Member of a SCRUM team, Developed path planning algorithm for selective laser sintering with concurrent laser,
reducing the print time by an average of 20%.
- Debugged legacy code that handle laser steering as well as library of geometry 2D and 3D in C++ and C#.
- Experimented with CGAL and other libraries that compute boolean operation on 3D meshes.
09/2013-02/2017 PhD
PhD Student, supervised by Sylvain Lefebvre
- Developed a graph based interface (github) in C++ using ImGui, to simplifies constructive solid geometry for 3D
printing. The interface is a mod of IceSL, a Slicer available here, developed by INRIA.
- Developed a path planning algorithm that improves the visual quality of 3d printed object.
- Developed an algorithm that generates support for 3D printing, reducing the printing time and the amount of wasted
plastic by up to 50%.
- Developed an algorithm that simplifies 2D mechanical modelling for 3D printing, allowing to 3D print designs done
with simple mechanical modelling tool.
- Developed an algorithm that optimizes packing and the shape of laser-cut furniture, reducing by up to 18% the wood
wasted by the fabrication of the furniture.
- Developed tools to visualize a 3D printer path, with openGL using real time rendering technique.
- Published in the best conferences in computer graphics: SIGGRAPH and Eurographics (Publications listed here).
- Teached algorithm and Programming(Java, Python, C#) and video games programming with Unity to
master/undergraduate student. Teached network programming to undergraduate student.
|
Skills
|
- C, C++, C#, Python
- Industry-grade library: CGAL, LibIgl, Tensorflow(Keras)
- 3D Printing: FDM, DLP, Inkjet, SLS
- Machine Learning: SVM, CNN, MLP
- English (fluent), French (native)
- OpenGL, GLSL
- Algorithm, Geometry
- Visual Studio, JIRA, Git, Nvidia Nsight Computer
|
Education
|
2017 PhD in Computer Science, Synthesis of Fabricable Shape from Partial Specifications,
INRIA Nancy, Université de Lorraine, defended in February 2017.
2013 Master degree in Computer Science, Machine Learning, Image Processing, Université
de Lorraine.
2011 Bachelor in Computer Science Université de Lorraine.
|