CV

Experience

AI Software Engineer @ Babylon Health June 2019 - Now

Making healthcare accessible and affordable for everyone - I work on the symptom checker.

Machine Learning Sabbatical: July 2018 - June 2019

• Deep Learning Paper Implementations:
  • I implemented a modified version of the decoder component from Vaswani et al. Attention Is All You Need and used it to train a language model with a baking cookbook.
  • I implemented the 1cycle learning rate policy from Leslie N. Smith. Super-Convergence: Very Fast Training of Neural Networks Using Large Learning Rates and verified the reported empirical results.
  • I implemented the learning rate range test from Leslie N. Smith. Cyclical Learning Rates for Training Neural Networks and verified the reported empirical results.
• Machine Learning Projects:
  • I trained a character level language model using an LSTM on the BNF and wrote inference code to run it in the browser for a fun(-ish) mini-game.
  • I entered the Human Protein Atlas Image Classification Kaggle competition with a friend and finished in the top 22% (Team Name: Eigencorp), despite not being able to afford sufficient GPU time to train on the full resolution images :(
  • I entered South East Water's preventing bursts hackathon with a friend, our solution got a special mention from the judges and we were awarded with a small cash prize.
  • I entered the House Prices: Advanced Regression Techniques Kaggle competition and got a score in the top 20%.
  • I did a detailed EDA on the data from the West Nile Virus Prediction Kaggle competition.
• Courses:
  • I read Tibshirani and Hastie's Introduction to Statistical Learning With applications in R and did the labs and exercises in Python rather than R. I implemented several of the algorithms from that book and The Elements of Statistical Learning from scratch in numpy.
  • I took Andrew Ng's Machine Learning, Neural Networks and Deep Learning and Sequence Models courses. I also took Jeremy Howard's fast.ai course.
  • I took Pavel Grinfeld's Linear Algebra course from Drexel, available via lem.ma and took Imperial College's Mathematics for Machine Learning specialisation.

I kept a learning log from 15/07/2018 up until 18/12/2018, when I got tired of keeping it up to date :s

Software Engineer @ Touchpress (which became) Amphio: July 2015 - July 2018

• Systems: Our team delivered core services for important artistic and scientific institutions. I was lead software engineer on the ISO Library and IOP events projects. I wrote the web front-ends in Clojurescript and implemented (AWS) lambda functions and workers on the backend in Python. I deployed these systems on AWS using Terraform and used CircleCI for continuous integration and delivery.

• Prototyping: I implemented a novel distributed PA system called "Soundarama" for a series of events at Frank Gehry's iconic Fondation Louis Vuitton gallery in Paris. The system connected one hundred iOS devices owned by the gallery to enable a DJ, using an iPad, to play audio stems though the devices in very tight musical time. Participants could augment the sound using the accelerometer and gyroscope in their devices. I wrote the applications in Swift using the functional reactive programming paradigm afforded by the ReactiveCocoa library.

• New Platforms: We had a great relationship with Apple, Tim Cook visited us during a trip to London. We were provided with pre-release access to Apple devices, including the AppleTV, iPad Pro and Apple Pencil. I implemented major components of the first applications for these devices including Classical Music Reimagined and Millie Marotta's Colouring Adventures.

• Workflows: I introduced several tools to improve our internal workflows:
  • I wrote scripts to safely migrate repositories from our legacy SVN system to git.
  • I wrote a small web application which our video production team's clients used to stream the latest work.
  • I wrote slack integrations which delivered notifications when AWS alarms were triggered in our production systems.
  • I introduced Terraform and championed auditable and reproducible infrastructure in AWS.
  • I introduced Fastlane for automated distribution of iOS builds to TestFlight and the App Store.

Core Contributor @ ResearchKit: Spring 2015

I was thanked by Apple for my contributions during a presentation at Apple's Worldwide Developers Conference. My contributions are used in production by organisations leveraging ResearchKit to undertake medical research, they include:

• A cognitive assessment based on the Tower of Hanoi #318.
• A reaction time assessment, using the device's accelerometer #163.
• Contributions to the libraries for rendering graphs #178.

Other Experience

• Software Engineer (Contracting) @ Fueled (NYC): Summer 2013
  • I was contracted to work on a novel iOS messaging application called Cooli.
• Software Engineer (Intern) @ Grapple Mobile (London): Summer 2012
  • I was selected to participate in the competitive Grapple Summer Academy. Working in team alongside other participants from France and India I developed a prototype mobile app for a London bar and lifestyle brand.
• International Representative @ MyCube (Singapore): Summer 2011
  • I was selected from over 15,000 international prospective candidates to complete this summer residency at a digital privacy start-up.

Other Open Source

• I implemented a multilayer perceptron from scratch in Clojure and used it to classify hand drawn digits in the browser (2017).
• I fixed a small issue that bit me in the fast.ai library (2018).

Education

• Bristol Medical School September 2008 - August 2014
  • Bachelor of Medicine, Bachelor of Surgery (MbChB)
• Dulwich College, September 2001 - June 2008
  • A-Levels: 3A's
  • AS-Level: 1A
  • GCSEs: 7A*'s 3A's

References available upon request.