Graham Hesketh


Dr Graham Hesketh, as a Data Scientist at Trilateral Research, is involved in the design and development of software and data science tools for public and private clients. He is active in all aspects of data curation and management across the data lifecycle and contributes to project proposals, research, and report writing in data analysis, data visualization, cloud computing, big data and machine learning.

As part of the EU funded big data project PROTEUS (Horizon 2020, Ref: 687691), Graham is currently leading Trilateral’s design and monitoring of KPIs and benchmarks. PROTEUS is a machine learning big data platform built on top of Apache Flink, accompanied with a custom-built javascript plotting library for streaming data and applied in a steel industry use case.

At Trilateral, he has designed and built, among other analytics tools, a machine learning web app that predicts air pollution from noise pollution collected on a smartphone, a deep learning entity resolution tool for database management and a research collaboration network visualization system built on the OpenAIRE database.

Graham is currently leading a small team of data scientists building the data analytics and cloud solution architecture in Trilateral’s STRIAD software for strategic risk assessment and data-driven decision making in law enforcement. This includes the front end development of the visualizations (html, javascript, css), the backend built in Amazon Web Services (Lambda, DynamoDB) and the security, web hosting and user profiles (Cognito, S3, Route53, IAM).

Other personal projects include social media analysis with Twitter (Clustering, topic/sentiment analysis), an electoral swing predictor (Gradient Boosted Regression), a probability to vote predictor (Logistic Regression), a house price predictor (Random Forests), and a likely party to vote for predictor (Neural Network).



Graham has authored and co-authored peer-reviewed journal articles, acted as a reviewer for leading journals, presented at conferences across the world, recently hosted the deep learning session at the big data conference STRATA 2018 and has over eight years’ experience in computational research, applied statistics and data analytics.

His background research involved large supercomputer simulations of optical phenomena including long-haul fibre optic communications, nonlinear optical processes and digital signal processing.

He was awarded the EPSRC doctoral prize for research excellence in 2014.

Graham holds a BSc. in Physics with Astrophysics from the University of Kent, an MSc. in Quantum Field Theory from Imperial College London and a PhD in the Computational Physics of Optical Communications from the University of Southampton, where he has also conducted postdoctoral research.