Machine Learning - Computer Vision - Computer Graphics - Physical Simulation

About Me

I am senior researcher at Microsoft, where I work on the technology that powers the next generation of digital content in virtual and augmented reality.

My work lies at the intersection of machine learning, computer vision and graphics. Please see google scholar for a complete list of my academic work.

Expressing ideas in code has been a great passion throughout my life, and I particularly enjoy optimising that code to maximum efficiency so it can run outside of an academic setting.

Apart from machine learning, I have a detailed understanding of computer graphics, imaging, data processing, computational physics, and parallel computing. I am proficient in Python, C, C++, and CUDA.


Real-time Neural Radiance Fields

My paper "FastNeRF High-Fidelity Neural Rendering at 200FPS", one of the first truly real-time NeRF models, was presented as an oral at ICCV 2021. Please check the link for further info, including explanatory videos. This is joint work with Marek Kowalski.

Appearance Transfer

My paper "High Resolution Zero-Shot Domain Adaptation of Synthetically Rendered Face Images" was presented and published at ECCV 2020. In this work, I developed a novel method that makes synthetic renders of faces at 1K resolution more believable. Please see the paper for more detail, especially regarding how bias is mitigated.

Appearance Synthesis

I am second author on the paper "CONFIG: Controllable Neural Face Image Generation", published at ECCV 2020. In this work, we use a novel model utilising adversarial learning to create an image generation framework that can be controlled via interpretable inputs. Code is available here.

Eye Tracking

As an intern at Facebook, I created the first version of the OpenEDS dataset, as described in our paper, published at ETRA. The annual conference workshop based on this work is still active today. I also worked on machine learning and graphics research during this time.

Interpretable Feature Space Transformations

I am second author on the paper "Interpretable Transformations with Encoder-Decoder Networks" (published at ICCV 2017). In this work, we demonstrate how to constrain feature space rotations to correspond to changes in images in autoencoder frameworks.

Harmonic Convolutions

I am second author on the paper "Harmonic Networks: Deep Translation and Rotation Equivariance" (published at CVPR 2017). The complete codebase for both the core algorithms and tensorflow training environment can be found here.

Interactive Computer Graphics WebGL Framework

UCL contracted me in 2016 to create and design an interactive, real-time coursework framework. Using webgl, complex algorithms can be implemented and debugged in real-time. The result can be found at cg.cs.ucl.ac.uk.

University Tech Talks

I have given a number of talks during my time at university, most notably:


Harmonic Convolutions

I have written a large part of the code for the paper "Harmonic Networks: Deep Translation and Rotation Equivariance". The complete codebase for both the core algorithms and tensorflow training environment can be found here and has received positive feedback since its release.

Denoising Framework

I have been interested in advanced image denoising algorithms since working in this area for the Moving Picture Company, London. This repository contains my general c++ denoising framework, including a full implementation of the BM3D and Non-Local Bayes algorithms. It also includes code for making those algorithms interact with neural networks and is fully multi-threaded using boost.

CUDA-accelerated Position-Based-Dynamics for Viscoelastic Tissue

The main project of my second MSc project dealt with the simulation of viscoelastic tissue for surgical simulation. I was awarded high distinction marks for my thesis "Simulating Organic Tissue with Position Based Dynamics" under the supervision of Dr Danail Stoyanov. The code for this project can be found here. It contains classes for advanced data IO of deforming 3d meshes using alembic, and a solver capable of simulating 20000 tetrahedra at 200FPS on a Geforce GTX 680.

CV & Experience


This is a short list of my academic history.

  • University College London: PhD Computer Science (2016-2022)
  • University College London: MSc Computer Graphics, Vision & Imaging, (First Class Honours) Distinction (2014 – 2015)
  • University College London: MSc Computer Science, (First Class Honours) Distinction (2013 – 2014)
  • Kings College London: Bachelor of Laws (LLB), Law 2:1 (2010 – 2013)


I have spent the last two years working at Microsoft as a researcher in machine learning / graphics / imaging. Prior to that, I worked as a research intern at Facebook and Microsoft Research. During my Master's, I worked as an intern and then a contractor at the Moving Picture Company in London. Please see my LinkedIn for further details.


I am an expert in deep learning, as well as more classical machine learning approaches. In particular, I love to combine neural networks with priors and constraints from physics, e.g. light transport models as demonstrated in the FastNeRF paper.

I speak:

  • English (native proficiency)
  • German (native proficiency)

I am proficient in the following programming languages:

  • c
  • c++ (98-14)
  • CUDA
  • python
  • c#
  • javascript (+html, css)

I have additionally got good knowledge in the following:

  • Haskell
  • VEX (SideFX Houdini)

Libraries I have worked with extensively:

  • PyTorch
  • Numpy/Scipy
  • Tensorflow
  • Matlab
  • Alembic / OpenEXR
  • Intel Threading Building Blocks (TBB)
  • Eigen
  • Intel Math Kernel Library
  • CUDA libraries such as Optix

Contact Me

If you any questions or would like to get in touch, please feel free to send me a message on LinkedIn, or drop me an email at stephan.garbin.13 at ucl.ac.uk, or alternatively myFirstName+myLastName@outlook.com