Experience

Hudl

• United States
• Software Development
• 700 & Above Employee

• Data Scientist

  • Jun 2022 - Present



University of Oxford

• United Kingdom
• Research Services
• 700 & Above Employee

• Data Scientist

  • Oct 2019 - Jun 2022

  Co-founder and acting Chief Technology Officer of AI diagnostics health-tech spin-out:• Led pitch to secure Lab10x start-up accelerator funding (250k)• Designed novel, multi-stage algorithms fundamental to the business’ core product, utilising object detection, image classification and Bayesian inference (patent pending)• Chiefly responsible for business codebase, encompassing data science, machine learning lifecycle, and web application capabilities• Developed data analysis code both independently and collaboratively to medical device regulatory standards, using agile methodologies and test-driven development• Prototyped and deployed deep learning algorithms to Azure ML cloud infrastructure, using continuous integration principles, and tools such as Docker and Kubernetes• Built Android application, using Kotlin and Java, to demonstrate key user journeys and prove the viability of deploying ML algorithms on a mobile device.• Collaborated with consultants to integrate MVP development with a quality management system suitable for a medical device, with the goal of attaining CE marking and regulatory approval• Managed full experimental data processing pipeline, including SQL data storage, data visualisation and analysis, as well as supporting the design of clinical data collection and experimentation protocols Show less

• Researcher DPhil Student

  • Oct 2015 - Oct 2019

  My DPhil, jointly supervised by Prof. Andrew Zisserman and Dr. Christoffer Nellåker, was based in the the field of computational syndrome diagnosis, with the title: "Aiding Diagnosis of Rare Diseases from Photographs using Machine Learning".A key goal of this area of research is to shorten the diagnostic journey for patients with rare genetic disorders, allowing them to receive treatment sooner. This could be accomplished by analysing ordinary facial photographs of patients using computer vision algorithms to narrow the diagnostic search space.The primary aim of my research was to build a new generation of patient datasets using facial images of patients with rare genetic disorders from the published biomedical literature. Secondly, I worked to quantify the effects of inheritance on facial features, so that these signals could eventually be disentangled from the facial features caused by an underlying rare genetic disorder. Mining rare genetic disorder patient information from biomedical literature:• Created and documented software pipeline to extract patient data from research articles• Trained deep convolutional neural networks to classify research figures containing faces• Developed website solution to enable fast annotation of large-scale retrieved dataModelling facial kinship using computer vision:• Trained novel, kinship-blind, network to benchmark key bias in current kinship data sets• Created largest kinship dataset by mining open-source ancestry data. The cleaned dataset contains the faces of over 145k individual people from more than 25k families• Adapted GhostVLAD network architecture to model facial kinship using feature aggregation Show less

• Systems Biology DTC Student

  • Oct 2014 - Oct 2015

  • Received six months of intensive training in biomedical data science and research skills • Completed two research projects including Patient Time Series Analysis for Mental Health Management



MBDA

• Defense and Space Manufacturing
• 700 & Above Employee

• WSSE Summer Intern

  • Jul 2012 - Sep 2012

  Participated for a second year in the summer internship programme at MBDA as part of the Weapon Systems Simulation & Experimentation (WSSE) department: • Designed target/tracking-window intersection algorithm for simulation software • Tested physics simulation software against current state-of-the-art • Developed a calculator plug-in for use with departmental proprietary software • Managed full diagnostic and mapping of the department’s hard-wired networks Participated for a second year in the summer internship programme at MBDA as part of the Weapon Systems Simulation & Experimentation (WSSE) department: • Designed target/tracking-window intersection algorithm for simulation software • Tested physics simulation software against current state-of-the-art • Developed a calculator plug-in for use with departmental proprietary software • Managed full diagnostic and mapping of the department’s hard-wired networks



MBDA

• Defense and Space Manufacturing
• 700 & Above Employee

• Facilities Summer Intern

  • Jun 2011 - Sep 2011

  Participated in the summer internship programme at MBDA as part of the Facilities department: • Audited conferencing capabilities to improve efficiency and reduce cost • Directed the tendering process for a large-scale conferencing projection system Participated in the summer internship programme at MBDA as part of the Facilities department: • Audited conferencing capabilities to improve efficiency and reduce cost • Directed the tendering process for a large-scale conferencing projection system