MEng Information & Computer Engineering
Cambridge University 2018 - 2019
1st class with Distinction (second place in cohort)
Taught aspects focussed predominately on machine learning, information theory and numerical optimization.
Master's thesis under Dr. Hernandez-Lobato revolved around adapting transformers for de-novo generative drug design.

BA Information & Computer Engineering
Cambridge University 2015 - 2018
1st class (second place in cohort)
Overall received 3 awards for academic distinction and 2 for programming exercises/competitions.

The first two years were very broad, including civil & electrical engineering, fluid mechanics, information theory and naturally a lot of mathematics. I later specialised in information and electrical engineering, with modules in computer graphics, medical imaging, digital electronics, statistical signal processing and probabilistic inference.

I also did many things on the side, including founding the robotics society, ballroom and hiphop dance, and a musical at one point.

Professional experience

Research Software Engineer
Google | Zürich, Switzerland
May 2021 - present
Continued computer vision research at Google Brain, focussing on scale, sparse mixtures of experts, discriminative & generative multimodal modelling, and medical imaging.

AI Resident
Google | Zürich, Switzerland
Oct 2019 - May 2021
Joined the Google Brain team, working predominately on computer vision, with focus on transfer learning, mixture of expert models, and applications to medical imaging.

Machine Learning Research Intern
ARM | Cambridge, UK
Summer 2018
Worked on improving efficiency of ML inference on the edge, with two main threads: extreme quantization of neural networks, and experimental efforts surrounding sparsity and Winograd convolutions.

Took part in - and won - their yearly Global Intern Innovation Competition, for which we developed a program for planning health facilities in Colombia (github).

AMGEN Research Scholar
ETH | Zürich, Switzerland
Summer 2017
Worked on medical imaging segmentation under weak supervision with Dr. Christian Baumgartner and Professor Ender Konukoglu. I combined classical computer vision approaches with machine learning to train neural networks on partially annotated semantic segmentations, which also involved developing a fun GUI for reannotating data. This work was continued by a Master's student and resulted in a paper at DLMIA. 

Various skills

Programming: Predominately Python as of late. Various experiences with C++, MATLAB, Java, SQL & VBA.

Machine learning: Fluent user of JAX, TensorFlow (1 and 2) and Keras.

Graphics design: Over a decade of experience with Adobe Photoshop & inDesign, alongside the usual MS office-esque applications.

Linguistic: Native English, ~B.2 German, intermediate Arabic.

Vehicular: I am legally allowed to drive but too afraid to get in a car.