Work
2024–now
CTO and co-founder
Lorefully
Co-founded the company and lead its engineering. Built the platform for capturing and structuring expert knowledge, and the evaluation environment that tests which models answer specialist questions correctly. It currently runs a UK-wide analysis of shipbuilding skills for the National Shipbuilding Office.
2018–2023
Head of Software and Data
Aceleron
Led software for energy storage products, setting the data strategy and applying machine learning to battery optimisation. Built the connected devices and cloud platform behind the BATLAB data services, deployed in Uganda, Kenya, Barbados and the UK.
2016–2018
Postdoctoral researcher
SUNCAT Center, Stanford University
Combined machine learning with genetic algorithms to search for new catalyst materials, cutting the expensive calculations needed by a factor of 50. Led development of the group’s machine learning package for surface science and catalysis.
2014–2016
Postdoctoral researcher
Technical University of Denmark
Searched for better fuel cell catalysts through atomic-scale simulation. Contributed a genetic algorithm to widely used open-source simulation software, extending it to find Pareto-optimal trade-offs.
2009–2014
PhD, Hydrogen and Fuel Cell Systems
University of Birmingham
Modelled platinum-alloy nanoparticles as fuel cell catalysts, combining genetic algorithms with quantum-mechanical calculations on national supercomputers.
2008–2009
MSc Molecular Modelling
Cardiff University
Simulation methods across protein folding, quantum mechanics and crystal structures.
2005–2008
BSc Forensic Science
University of Glamorgan
A computer forensics module is where the programming started.
Selected publications
2019301 citations
Genetic algorithms for computational materials discovery accelerated by machine learning
P. C. Jennings, S. Lysgaard, J. S. Hummelshøj, T. Vegge, T. Bligaard. npj Computational Materials 5, 46.
Training a machine learning model on the fly cut the expensive energy calculations a genetic algorithm needs by a factor of 50.
2019257 citations
Low-scaling algorithm for nudged elastic band calculations using a surrogate machine learning model
J. A. Garrido Torres, P. C. Jennings, M. H. Hansen, J. R. Boes, T. Bligaard. Physical Review Letters 122, 156001.
A machine learning stand-in for expensive quantum calculations when mapping the path and energy barrier of a chemical reaction.
20176,744 citations
The atomic simulation environment: a Python library for working with atoms
A. Hjorth Larsen, J. J. Mortensen and others, including P. C. Jennings. Journal of Physics: Condensed Matter 29, 273002.
The reference paper for the open-source simulation package that carries the genetic algorithm I worked on at DTU.
201490 citations
A DFT study of oxygen dissociation on platinum based nanoparticles
P. C. Jennings, H. A. Aleksandrov, K. M. Neyman, R. L. Johnston. Nanoscale 6, 1153–1165.
From my PhD: how oxygen splits apart on platinum nanoparticles, the first step of the reaction that hydrogen fuel cells depend on.
Contact
I’m glad to hear from people working on machine learning evaluation, on turning research into products, or on data in hardware businesses. I reply to anything specific.
Profiles