Jonathan Deakin

Bio

As an applied mathematics PhD student and technical co-founder at Etymo, I am motivated by finding and communicating powerful mathematical solutions to everyday problems. I have experience in scientific computing, writing efficient software in high level (Julia, Python, MATLAB) and low level (C++) languages. I am also interested in AI, web technologies, databases and computer hardware.

Education

Currently studying for an applied mathematics PhD at the University of Manchester. Previously I studied Maths at Jesus College, Cambridge. Coursed taken include Numerical Analysis, Fluid Dynamics, Optimization and Control, Statistical Modelling, Mathematical Biology, Computational coursework using MATLAB, Deep Neural Networks and Dimensionality Reduction.

Projects

PhD

The title of my PhD project is: "Efficient numerical solvers for acoustic fluid-structure interaction problems" and is supervised by Professor Matthias Heil and Professor Andrew Hazel.

I am an active contributor to oomph-lib, a C++ library for solving partial differential equations using the finite element method. All machinery developed during my PhD has been implemented in oomph-lib. I am also nominally responsible for the styling of the website too.

I co-created MultipleScattering.jl, which is a Julia library for simulating the acoustic scattering from multiple particles. The library uses multiple dispatch instead of object oriented principles to build general machinery which could be easily extended to different numbers of dimensions, different equations (like the Poisson equation or linear elasticity), particle shapes and sources.

etymo.io

I am a co-founder and engineer at etymo.io, where we build knowledge graphs for enterprise. For an example, we developed scholar.etymo.io a search engine for AI researchers. More technical details can be found in our pre-print on arXiv explaining the mathematical underpinnings and system architecture of the search engine.

Arduino

The idea of a small, programmable device is as appealing to me as it seems to be to everybody else (see IoT and Raspberry Pi ). I started to play around with Arduinos during a summer project at PCMend.net and I've been hooked ever since.

I have lead the software development of a data-logger and pass-through device. The data logger was used to monitor water flow for a UK gym chain. The pass-through device passes messages from Ethernet to RS232, then sends back any replies. The unit can be configured by a webpage, reports its presence over Control4 SDDP and we are in the process of developing support for cloud control such as Alexa. It is currently in production and has been installed in commercial TVs.

Manchester SIAM-IMA Student Chapter

As the Webmaster for the chapter, it was my responsibility to maintain and update the website. This included redesigning the interface of the website, as the old interface had become dated and unsuitable for mobiles.

As a member of the committee, it was also my responsibility to organise conferences:

I recently ran a short seminar/workshop on basic web design skills for students at The University of Manchester. The content of the workshop can be found on GitHub.

Mondohack III: Mondonate

Inspired by our interest in transparent and effective charity, Charlie Whittaker and I attended Mondohack III. Organised by the startup bank Monzo (previously called Mondo), it was an invitation to play with their API. Our idea was that when a user donated money to a charity, we would detect this and give them information on what the donation would acheive. Thanks to the admirable efforts by charities like the Against Malaria Foundation, we could then provide a donor with an update on its true impact right in their bank statement. Charity can seem distant, and as a result donors may give little thought to the actual impact. Tools like this could close the gap and encourage donors to consider their impact, perhaps choosing higher impact charities.

Several great people joined the team to contribute, and the product we presented (mainly written by Joe O'Brien) won a runner up prize. The code can be found on the Mondonate team GitHub.