Konstantinos Rematas
Staff Research Scientist at Google DeepMind
Seattle, Washington, Switzerland
Summary
👤
Senior🎓
Top SchoolKonstantinos Rematas is a Staff Research Scientist at Google DeepMind with a decade of experience at the intersection of computer vision, graphics, and generative AI, progressing through research roles at Google and academic positions including a PhD from KU Leuven. He has driven foundational work in neural rendering and volumetric techniques—contributing implementations like trilinear interpolation, positional encodings, and NeRF-style volumetric rendering to the widely used TensorFlow Graphics ecosystem. Based in Seattle with roots in Europe, he blends rigorous academic research with production-minded engineering, shipping research that bridges differentiable graphics and scalable ML systems. Known for synthesizing novel views and realistic video generation, he brings deep expertise in 3D sampling and rendering that feeds into current generative video efforts at DeepMind.
10 years of coding experience
13 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Computer Vision and Graphics, Doctor of Philosophy (Ph.D.) Computer Vision and Graphics at KU LeuvenBSc Computer Science, BSc Computer Science at Aristotle University of Thessaloniki (AUTH)MSc Media Informatics, MSc Media Informatics at RWTH Aachen University
Github Skills (5)
raytracing10
machine-learning10
tensorflow10
computer-graphics10
python9
Programming languages (1)
Github contributions (5)
tensorflow/graphics
Feb 2020 - Oct 2021
TensorFlow Graphics: Differentiable Graphics Layers for TensorFlow
Role in this project:
ML Engineer
ML EngineerContributions:30 commits in 1 year 9 months
Contributions summary:Konstantinos contributed significantly to the TensorFlow Graphics library by implementing and testing trilinear interpolation for regular 3D grids, a foundational element for volume sampling, and also added and tested various sampling techniques for rays. They also introduced and tested the NeRF model and implemented the positional encoding functionalities which are essential for the performance of the NeRF. Further contributions involved the implementation of a volumetric rendering approach.
differentiablecomputer-graphicslayerstensorflowgraphics
krematas/soccerontable
Apr 2018 - Dec 2018
Upconverting YouTube soccer videos in 3D for viewing in AR/VR devices.Soccer On Your Tabletop
Contributions:57 commits, 44 pushes, 2 branches in 7 months
virtual-realityvideosar-vryoutubesoccer
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.