Data Scientist
CurrentDesign and maintain relational databases to ingest realtime data from a number of data sources, delivered in varying formats such as XML, JSON, websocket, as well as live GPS data through UDP.Maintain public facing REST API and website for racecourse pars, horse ratings, graphical visualisers.Maintain an nginx load balancer to direct and balance traffic to various different backend services, some of which are described above, deployed in AWS EC2 and RDS instances.Identify and remove trackers that are performing significantly poorly compared to others, using statistically derived confidence intervals of errors filtered using a plotly/javascript frontend of my development.Model horse performance against several kinds of proprietary racecourse pars using attributes derived from GPS points and stride data, used to generate ratings and insights to drive customer and media interest.Improve the inrunning race pricing model utilising both Neural Networks (TensorFlow) and traditional statistical analysis. Improvements apply to both the computational efficiency of the deployment, and the theoretical profit/loss of the model prices, scored using an innovative method of my design which has shown to well reflect the actual P/L observed by customers.Customer point of contact for general questions about the data, network and code troubleshooting, recommended software etc.
