London, England, United Kingdom
Machine learning researcher with a PhD in particle physics and a range of experience in research and business environments. Looking for opportunities to develop big-data and machine learning solutions which deliver impactful results.
- PhD in experimental particle physics. Working as part of the ATLAS Collaboration, a large international team of world-leading scientists and engineers. - Included period of Long Term Attachment working at CERN in Geneva, Switzerland. - Working with terabyte-scale datasets from simulations and particle collisions. - Using cutting-edge statistical techniques to perform data analysis, including regression and classification machine learning algorithms such as boosted decision trees and neural networks. - Lead author on academic paper targeting publication in an international physics journal. Research is focused on measuring the phenomenon of CP-violation - why the laws of physics are different for matter and antimatter. Technologies used: Python, Scikit-learn, TensorFlow, Keras, C++, ROOT, GitHub. Techniques used: Statistics, machine learning, deep learning, boosted decision trees, gradient boosting, unfolding/deconvolution, Monte Carlo simulations, systematic & statistical errors.
- Involved in experimental operations and upgrade of the ATLAS trigger system. - Redesigned the trigger configuration database, achieving speed-up of 100-1000x in editing and loading configurations. - Implemented Python codebase and command-line interface to manage all database interactions. Technologies used: Python, SQL, Oracle, cx_Oracle, GitHub. Techniques used: Relational databases, query optimisation, cryptographic hashing.
- Teaching assistant for: 1st year undergraduate course in Mathematics for Scientists, leading tutorial classes on calculus, linear algebra, probability, and statistics; 2nd year undergraduate course in Python programming. - Marking assistant for 1st year undergraduate course in: Mathematics for Scientists; 3rd/4th year undergraduate courses in Particle Physics, Climate Physics, and General Relativity; postgraduate course in Statistics for Data Analysis. Techniques used: Mathematics, probability, statistics, teaching.
- Internship within the Data Science & AI team, working on a state-of-the-art algorithm for learn-to-rank problems, built with TensorFlow/Keras deep neural networks. - Project was to implement knowledge distillation in order to reduce model size and improve live inference performance. - Independently completed literature search, design, code implementation, and performance testing. - Delivered solution that reduced model size by 70% without any loss of performance. - Prepared solution for production, planned future move to open-source. - Gained experience & training with big-data pipelines using Spark & Databricks. Technologies used: Python, TensorFlow, Keras, PySpark, Spark, Databricks, GitHub. Techniques used: Machine learning, deep learning, neural networks, knowledge distillation, model compression.
- Four-year undergraduate master's degree in Mathematics & Physics. - Final dissertation on 'Machine Learning for Higgs Boson Measurement'. Wrote a from-scratch Python implementation of the AdaBoost algorithm, with comparable performance to scikit-learn. Achieved 90%+ correct classification of signal and background. Technologies used: Python, Scikit-learn, Matplotlib. Techniques used: Machine learning, boosted decision trees, AdaBoost.
- Research internship in Geochemistry department at Durham University. - Used molecular physics simulation software to investigate the properties of glyphosate, a commonly used herbicide. Techniques used: Modelling, simulation.