Tasks:
- * Build automated multi-view data processing pipelines
- * Design cross-view geometric alignment and pixel-aligned feature prediction networks
- * Develop feed-forward 3D reconstruction algorithms
- * Generate synthetic 3D assets using Blender or simulators
- * Improve Gaussian initialization and densification strategies
- * Optimize 3D Gaussian Splatting reconstruction pipelines
- * Optimize CUDA and Triton rasterization kernels
Perks/Benefits:
Skills/Tech stack required:
[3D Gaussian] [3D Gaussian Splatting] [Blender] [C++] [Camera Calibration] [Cross-Attention] [CUDA] [Data-parallel] [DeepSpeed] [Distributed data] [Distributed Data Parallel] [Epipolar geometry] [Gaussian Splatting] [Linux] [Multi-view Geometry] [Multi-view Stereo] [Open3D] [Perspective projection] [Python] [PyTorch] [Pytorch3D] [Rasterization] [Structure from Motion] [Transformer Architectures] [Trimesh] [Triton]
Educational requirements:
[Bachelor's Degree] [Master's Degree] [PhD]
Role(s):
[3D Reconstruction Engineer] [Computer Vision Engineer] [Deep Learning Engineer] [Engineer] [Learning Engineer] [Vision Engineer]