Developing computer vision and data analysis algorithms for recognising and tracking cars from a traffic camera feed between sites. Developing Dual-Tree Complex Wavelet Transform based algorithms to perform real-time analysis of video.
Part of the Data Analysis for the LoLa single-molecule imaging team. Developed a custom suite of data analysis and visualisation tools for biological imaging and tracking. Created novel image and signal processing algorithms for dealing with noisy and ambiguous input. Co-ordinated geographically disparate team and lead efforts to improve code quality with automated build testing tools, test suites and coding standards.
Based at the University of Cambridge. Technology lead and inventor of the Pneumascan passive lung function monitor system. Rôle required investigation and development of computer vision algorithms for real-time high-frequency 3D data capture. Implemented a prototype capture and analysis system including GPU accelerated processing via CUDA. Created product mock up and marketing renders using the Blender rendering suite.
Researched new algorithms for rigging and skinning 3D character models resulting in the publication of a fully-automated system called Bone Glow
Real-time graphics research including full physical global illumination calculations performed via GPU and CPU co-operation. Developed a practical, real-world system for lighting now in use in well known game titles such as Battlefield 3 and Need for Speed: The Run.
I have a great deal of experience in various programming languages including C, C++, JavaScript and Python and have a passion for writing good, clean and well-tested code. In addition, I have an Academic background in Machine Learning, Computer Vision and Artificial Intelligence. I’ve contributed to a number of Open Source projects to greater or lesser degrees and my GitHub account https://github.com/rjw57 has a large number of repositories which give an overview of my areas of knowledge.
Open Source projects which I created and which are sufficiently widely used to be packaged in the popular Debian distribution of Linux.
This is written in C and was a project I started to provide support for DVD navigation in Open Source projects. At the time, I was a committer to the xine media player project and the library was created as a re-usable off-shoot from the DVD playback functionality I was adding. The library forms the core of the vast majority of Open Source DVD playback programs. In addition to parsing the somewhat baroque binary data structures used by the DVD Video format it also contains an implementation of the DVD Virtual Machine used to execute navigation programs. The library is now maintained by the mplayer project.
The dtcwt module for Python which is hosted at https://github.com/rjw57/dtcwt. It implements a particular technique in signal processing known as the Dual-Tree Complex Wavelet Transform (DT-CWT). This is used in computer vision, image processing and financial time series analysis. The library has multiple performant implementations of the numerically intensive DT-CWT algorithm for both CPU and GPU via OpenCL. I am mentoring a PhD student who is providing a further implementation using Tensorflow, a framework which is popular in the Machine Learning and Artificial Intelligence field.
Each pull request for the dtcwt library is tested via a Continuous Integration system which ensures correct functionality via a test suite and code coverage check. Documentation is automatically built and is available on the web. The library is structured so that it may be installed automatically using the standard Python pip tool. I passionately believe that software must be packaged, documented and tested well in order to be useful and I use the dtcwt module as an example to my students on how much quality assurance infrastructure is available in the cloud.
I am a confident public speaker and was invited to give the keynote talk at last year’s Diamond Developer Day. The event was a conference of all of the research software engineers for the UK’s largest particle accelerator. I gave a talk on software quality engineering in an Academic environment which was very well received.
I am passionate about the history of computing and the history of programming languages in particular. In order to learn more about compilers, I wrote a toy compiler for the B programming language which was the direct predecessor of the well-known C language. The compiler is available at https://github.com/rjw57/rbc and uses the powerful LLVM compiler infrastructure. It was intended as a teaching aid for compiler implementation and formed the core of a talk I gave last year at the Cambridge Python Users Group.
In order to learn electronics and to practice low-level and embedded programming, I decided to create my own computer from scratch. The result is the Búri microcomputer system (https://github.com/rjw57/buri) which is a 6502-based computer system with an Operating System written in 6502 assembler. The latest version of the hardware uses a 65816 CPU with 512K of memory and has an in-development video card.
As a more modern example, I co-designed and programmed an embedded inertial measurement unit as part of Cambridge University Spaceflight. This credit-card sized module integrated an STM32 microcontroller with various inertial, GPS and magnetic sensors to provide a real time estimate of spacecraft attitude and position. In essence it was a modern-day Apollo guidance computer.
I’ve tried to keep myself abreast of modern web development and am reasonably fluent in JavaScript. I have personal relationships with some people who work for the Portable Antiquities Scheme (PAS) and have created a couple of web apps related to their work which may be found at http://rjw57.github.io/findmyflo/ and http://rjw57.github.io/pas-measure/. I’ve contributed to the Open Source PAS database project. I’ve experience with React, WebPack and AngularJS all of which I’ve used for various internal work projects and some personal web app projects.
Researched applications of advanced mathematical geometry techniques for real-world use. Gained experience in geometric interpretation of signal processing techniques, knowledge of accelerated graphics APIs and developed new research in the fields of computational geometry, non-Euclidean problem solving and fractals. Thesis title: Computer Graphics using Conformal Geometric Algebra.
Received a distinction (class I equivalent) in Information Engineering. Course scope included signal processing techniques, computer graphics, artificial intelligence and software engineering.
I try to keep a broad range of technical skills up to date because, frankly, I like to play with technology and writing software. In general I am quick to ‘deep-dive’ into a new field and find it easy to pick up new programming languages and software-related skills. I find systems design extremely satisfying and am never happier than when all of the metaphorical plates are kept spinning by a robot which I have designed. I am particularly enthused by what has come to be called agile development.
I am a signal processor by trade which means that I am particularly interested in the inferences one may draw from data that one can observe. To this end I make heavy use of Bayesian methods such as Monte Carlo sampling in order to find maximally likely a posteriori estimates of the true state of a system from noisy observations of its output. My research interests are therefore wide-ranging and have wide applicability. My recent interests have been in fields as diverse as computer vision and analysis and inference of the dynamics of receptors on cell membranes. A list of my publications may be found on the University of Cambridge’s Department of Engineering website.
Extensive experience with C, C++, Python and various Unix glue languages such as sed, awk, perl and ed. Experience in writing for the Gtk+ and Qt GUI toolkits. I have written at least one non-trivial program in order to learn languages including Scheme, Smalltalk, Ruby, Java and C#. I class myself as a competent multi-paradigm programmer who is capable of picking up any sane house language or coding style with some rapidity.
Experience with HTML/CSS/JavaScript (including the latest HTML5 standards) over a wide variety of personal projects including, but not limited to, this CV, an interactive Anglo-Saxon dictionary, an Old English speech synthesiser, a client-side Google+ syndication system and webmaster for my group’s website. Developed a public-facing in-house course management system for the Department of Engineering which is used to manage third and fourth year module choices. Experience with server-side programming languages such as Python (WSGI), PHP, Perl (Mason) and Java servlets.
Enthusiastic student of realistic rendering techniques. Wrote an accelerated (via SSE) path-tracing renderer for Minecraft maps. Can program using both the OpenGL 2/3 and DirectX 9/10 APIs. Real-world experience of writing computer graphics software and GPU acceleration.
Member of the Many Core Computing Group in the University of Cambridge with a research interest in GPU acceleration of scientific computing. As part of the LoLa project, I have ported many of our compute-intensive algorithms to a multi-Tesla architecture and developed a computing engine allowing single source-code compilation with CUDA and OpenMP accelerated backends. Developed a novel multi-threaded Metropolis Hastings and Gibbs sampling algorithm to be published in a forthcoming paper.
I’ve been de facto system administrator in many places of work and although I would class myself as a knowledgeable amateur in this field I haven’t as yet lost anyone’s data despite many hardware failures over the years. I have created company-critical backup and restore systems along with designing a company source control and quality engineering work flow.
I enjoy languages and can speak French to a conversational level. I am currently studying Arabic, Anglo-Saxon and Welsh.
I have a full, clean UK driving license and have held it for more than ten years.
I enjoy typographic design work and have designed many posters for theatre groups in and around Cambridge. In addition, I have performed with many theatre groups and have taken several improvised comedy and comic theatre shows to the Edinburgh Fringe festival. I have long ago lost any fear of public speaking.