Refresh was a project conceptualised and developed within a week at the YRS Festival of Code 2015. The principal role of the project was to provide an easy, simple and intuative platform for users to revise for any exam, provided with an exam syllabus. We identified that the main problem with modern online learning environments is that the data driving them is either high quality but very low in quantity, or low quality and high in quantity, and so we devised a plan to utilise the extraodinary amounts of data available online at sources indexed on popular search engines, using this content to formulate challenging exam questions that would test the user's ability and develop their understanding. We then use data collected from user interactions to focus questions and revision time on subject areas which the user finds more challenging. We hope that this proof-of-concept demonstrates the ability of the world wide web as a data resource and encourage you to check out the project here.
After being invited to the Open Data Institute’s Summit 2015, we took part in a hackathon whereby we were briefed to develop a food based application that utilised open data in some form. Having spent a very short period of brainstorming, we settled on the idea to write a program to calculate the amount of time it would take exercising to burn off the calories consumed in a meal. To accomplish this, we utilised a dataset that contained the calorific content of hundreds of thousands of different foods, and set up a search field to select the food eaten. Upon entering a little bit of basic information such as weight, gender and age, we carry out a set of complex calculations to deduce the amount of time required swimming, running and walking to burn off the meal. In total, the project was fully designed and developed in under 4 hours.
Talos is a joint concept developed by Oli Callaghan and Finnian Anderson, that monitors the uptime of sites, and is able to alert the monitor when the site goes down. It utilises Docker to create and destroy site health monitors, written in NodeJS, whilst also utilising Docker containers for the storage of user data, serving front-end files, and providing an interface for Alexa to read live updates on site health. The application can be experimented and used at the link below, and the full source code can be found on Github.
Dendrite is a new concept developed for my Computer Science A-Level to allow users to train neural networks without requiring a single line of code to be written. Networks are constructed by dragging and dropping layers together, to construct a graph which can then be trained using user defined training data. Once trained, these networks can then be tested for their accuracy. The machine learning engine was written entirely from scratch using C++ and OpenCL, allowing the GPU to be used when executing networks, whilst the user interface was designed and developed using HTML, CSS, JS and Electron.
engine source code
ui source code
During the summer of 2016, I amongst a group of others were invited to work at BT Martlesham Heath in the customer showcase to develop new products and innovations to display for prospective investors. As they were currently renovating the bank showcase, I was involved in the refurbishment and redesign of a cash point within the premises that utilised a palm print recognition system to validate the identity of the user. In addition to this, I was part of the small team that was involved in developing a ground breaking new technology that allows phones hung by their charging cables to act in an array as one in a large screen. After taking a picture of the set up of phones, we utilised image recognition to find the positions of the phones, identify their locations, and then stream cropped parts of a video to them over Wi-Fi. The entire project was vast and extensible, providing a robust framework for advertisements, and can be found at the BT Customer Showcase at Martlesham Heath.