I was the Lead Data Scientist for Hitachi Social Innovation, tied to the Optimise Prime programme.

Optimise Prime is the worlds largest commercial electric vehicle trial. It is a £35 million Ofgem energy innovation project, with partners Uber, Royal Mail, Centrica (British Gas), the energy regulator Ofgem and Distribution Network Operators UKPN and SSEN.

The project helped commercial partners understand the best way to transition their petrol and diesel fleets to electric, from using data analysis to recommend appropriately ranged vans to machine learning to understand driver behavioural patterns and recommend chargepoint installation zones in London. It helped grid partners trial commercial level flexibility services, pivotal towards their DNO to DSO transition, and give different fleet types exposure towards the potential benefits that engaging in such services could bring them.

Notable Achievements

  • Managing team of 2-6 people (including internal and external contractors). Mentored 1 Junior Data Scientist. Interviewed candidates for Data Science positions.
  • Led data science efforts across three trials, with a combined total of over 6,000 electric vehicles, using agile delivery methodology.
  • Responsible for technical delivery of key deliverables, both external and internal, as well as models and algorithms. A technical talk on our methodology of a state-of-the-art geospatial machine learning model to predict chargepoint demand was given at Big Data LDN 2020.
  • Used data science techniques to build an explanable depot planning model which delivered over £1.8m in operational savings for Royal Mail across 10 depots while ensuring risk and cost minimsation. It calculated optimal grid connection agreements which can be used to avoid potential network reinforcement costs to the tune of £millions/year. This was deployed as a cloud web app, as an online public tool, which used modern web frameworks (Django on Azure), REST APIs and developed using good coding best practices.
  • Data analysis and visualisation to cluster behaviours into archetypes, useful for data storytelling and communicating to stakeholders.
  • Creation of simulation based tools, to enable partners to ask “what-if” questions.