Nikos Kolotouros
Machine Learning Engineer
Zurich, Zurich, Switzerland
Summary
🤩
Rockstar🎓
Top SchoolNikos Kolotouros is a Machine Learning Engineer and Research Scientist at Google DeepMind with eight years of experience building multimodal generative AI systems for video and pose control, following a PhD in Computer and Information Science from the University of Pennsylvania. He blends research rigor with production-minded engineering—having implemented core components like a PyTorch 3D mesh renderer with CUDA kernels and advanced training pipelines for 3D human pose and shape reconstruction (SPIN). Comfortable across model training, optimization, and low-level performance plumbing, he has a track record of stabilizing complex ML systems and integrating camera/SMPL fitting into robust training loops. Based in Zurich, he brings a rare combination of deep computer vision expertise and hands-on systems work that accelerates moving cutting-edge research into usable multimodal tools.
8 years of coding experience
University of Pennsylvania
French, English, Greek
Github Skills (12)
cuda10
computer-vision10
3d-rendering10
pytorch10
machine-learning10
loss-functions10
trainings10
python10
modeling10
optimization9
linear-algebra8
data-augmentation8
Programming languages (1)
Github contributions (5)
daniilidis-group/neural_renderer
Jun 2018 - May 2019
A PyTorch port of the Neural 3D Mesh Renderer
Role in this project:
Back-end Developer
Back-end DeveloperContributions:134 commits, 13 PRs, 29 pushes in 11 months
Contributions summary:Nikos implemented a PyTorch port of a 3D mesh renderer, initializing the project and porting key functionalities. They developed the `get_points_from_angles` function and integrated lighting calculations. Further, the user added CUDA kernels and wrappers for texture loading and implemented the `look_at` and `look` functions. They also translated the perspective transformation to PyTorch.
pytorchmeshrenderer3d-mesh
nkolot/SPIN
Sep 2019 - Apr 2020
Repository for the paper "Learning to Reconstruct 3D Human Pose and Shape via Model-fitting in the Loop"
Role in this project:ML Engineer
ML EngineerContributions:18 commits, 6 PRs, 15 pushes in 6 months
Contributions summary:Nikos contributed to the training code and made several modifications to the trainer.py file, including integrating datasets, models, and optimizers. They also focused on implementing loss functions and optimizing camera parameters and SMPL fitting processes, suggesting a core role in refining the model training pipeline. Furthermore, the user addressed model stability issues by adding checks for extreme beta values. These actions indicate a focus on debugging and improving the model training process.
pytorchhuman-posedeep-learningcomputer-graphicscomputer-vision
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.