Context
Metalograms is bridging the gap between cinema and gaming. We use 3D & 4D Gaussian Splatting (3DGS/4DGS) to bring photorealistic volumetric video to Virtual Reality. Our next milestone is to launch the development of an XR video game showcasing the potential of this technology.
We are looking for a Machine Learning / Computer Vision student passionate about cutting-edge 3D generative techniques to take ownership of our 3DGS/4DGS training, finetuning, and benchmarking pipeline.
Missions
As our Machine Learning & 3D Computer Vision Intern, you will bridge the gap between raw data capture and real-time rendering by taking full ownership of our upstream 3DGS/4DGS pipeline:
- Pipeline Automation: Design and automate the end-to-end data processing workflow, transforming raw multi-view video/image sequences from our cameras rig into finalized PyTorch-trained 4D Gaussian Splat assets via structure-from-motion (Colmap/SfM).
- SOTA Implementation: Research, integrate, and test recent State-of-the-Art 3DGS/4DGS papers to enhance visual fidelity, eliminate floating artifacts, minimize the number of gaussians, and improve the structural accuracy of dynamic 4D volumetric scenes.
- Model Optimization & Benchmarking: Profile training times and VRAM footprints. Implement compression, pruning, or quantization techniques to reduce file sizes, preparing the assets for mobile XR deployment constraints.
Requirements
- Education: M.S. or Master 2 student in Computer Science, Data Science, or Robotics, specializing in Machine Learning, Deep Learning, or 3D Computer Vision.
- Technical Core: Strong proficiency with PyTorch and Python data science ecosystems. Solid foundational understanding of 3D Computer Vision (camera matrices, stereo geometry, SfM/Colmap, NeRF, or Gaussian Splatting concepts).
- Bonus : Experience writing or modifying custom CUDA kernels or Vulkan Compute Shader
- Profile: High autonomy, self-learner, and deeply curious about the intersection of AI and real-time graphics.
Conditions
- Duration: 3 to 6 months. Starting Date: Automn
- Agreement: internship agreement needed with school or Erasmus+ agreement
- Compensation: ~661€ (31.50€ / day) + transport allowance
- Location: Bordeaux Center (startup coworking space)
Job Type: Internship
Contract length: 6 months
Pay: 661.00€ per month
Application Question(s):
- Avez-vous des projets personnels ou académiques en Machine Learning / Computer Vision ? Pouvez-vous fournir un lien GitHub ou un portfolio ?
Do you have any personal or academic projects in ML/CV? Can you provide a GitHub link or portfolio?
- Quelle est la durée et les dates souhaitées pour votre stage ?
What is the desired duration and starting dates for your internship?
- Quelle est votre expérience avec PyTorch ou la manipulation de données 3D (Colmap, nuages de points, NeRF, 3DGS) ? What is your experience with PyTorch or manipulating 3D data (Colmap, point clouds, NeRF, 3DGS)?
Work Location: In person
