I worked on six editions of the Statistical Review of World Energy (dedicating a third of each year to the effort). The review is downloaded more than 160,000 times every year from individuals in over 180 countries. It is the leading source of up-to-date, comprehensive data on the global energy system. Our data is available freely, for use by anyone, and has been cited by influential figures, like Bill Gates; global leaders in the media, like Bloomberg and the Financial Times; top-tier private firms, like Blackrock; and it is cited by government agencies around the world, including by the US EIA.The project was a collaborative effort between my team at the university and BP, which contracted us. They published the review for over 70 years. In 2023, it passed stewardship of the project to the Energy Institute. I have since worked with them, and with analysts from KPMG and Kearney, which provide sponsorship and substantial support.I led a workstream (to track the trade of fuels), and I delivered results on time every year.I led and collaborated with a cross-functional team; was responsible for data pipelines which collected, cleaned and aggregated information from over 100 sources; I administered the resulting data in a database using SQL; and I produced information and insights for publication.Additionally, I made several technical contributions to the project. I developed tools for processing information from surveys, which cut the analyst time required by 60%. I also proposed and implemented improvements to our processes for revising data, delivering efficiency gains on the order of 50%. This required architecture changes, training and new workflows.Finally, alongside colleagues, I have co-authored ad-hoc analysis on the global gas market, which has been published.

My prize-winning research was about technological change and its effects on key incentives governing communication. As part of the work, I assembled large, custom datasets and employed frontier methods in natural language processing (working with a high-performance computing cluster). In turn, that work enabled me to shed fresh light on results obtained from over 30,000 A/B tests, which had been conducted by a media company. I modelled the company's optimal policy as part of the research, and I demonstrated how their actions deviated from it.The extended duration of my research is due to my work on the Statistical Review of World Energy, which consumed a third of each year during the programme (and during which time, my research was formally paused).

I taught statistics and econometrics in a rigorous course from the core of the Scottish Graduate Programme in Economics (MSc/PhD). The material includes coverage of quasi-experimental methods, as well as machine learning.

I conducted a review of the company's machine learning models, which were designed to forecast asset failures and establish requirements for capital investment (~£500 million per year at the time). I was responsible for building support for the models to ensure they could command confidence – particularly from the firm's economic regulator. I delivered a number of recommendations for improvements, including some which were technical (relating to the nature of the models) and some which targeted the communication of their outputs. Many of these recommendations were subsequently adopted, and I assisted with their implementation.Separately, I was responsible for delivering estimates of the firm's costs that were disaggregated geographically. This was the first successful attempt to do so by a company in the sector. The work made it possible to pin down the degree of the returns to scale that underpin the firm's operating model, as well as its regulatory regime. It also enabled the firm and its regulator to better understand the trade-offs required for improvements to service delivery in rural areas – a priority for the regulator and government. Alongside the data, I developed a tool to help visualise, aggregate, and analyse it; I presented both to the regulator, and it was subsequently adopted for their internal analysis.Finally, working closely with the Director of Strategy, I created bespoke visualisations to better communicate the firm's investment strategy. The outputs highlighted mismatches between the strategy and the firm's objectives, resulting in adjustments worth tens of millions of pounds.

I worked on disability policy at the Department for Work and Pensions. I delivered statistical analysis which was published by the department, and I developed a new metric which was central to the evaluation of a manifesto policy.