Summary

Data Science Lead with 8+ years of experience delivering NLP and geospatial ML/AI systems for public sector and product companies in climate, urban intelligence and energy. I have led teams through data architecture, ML pipelines, modernisation and governance in production: £180k/year from energy pricing models, a 1,500-user agentic AI rollout, and the 100x speedup of a critical service scoring 30,000+ organisations a year. As a DS leader I uphold agile, DevOps and backend-engineering-grade coding standards. My training is complex systems physics (published in Physical Review E).

Experience

Senior Consultant - AI Applications

Solirius
Apr 2026 - Present, London

  • Saved the Department for Education ~£250k by automating validation of 23,000+ floorplans with a human in the loop ML validation service. Improved on Esri’s accuracy by 32% (58% to 90%).
  • Led 4 people enabling an agentic AI rollout to 1,500+ staff across 9.5M files at a national housing agency by designing the data governance approach. Approved by Governance, SecOps, the DPO, the Technical Design Authority and the Change Advisory Board.

Energy Data Science Product Lead

True (ex-Open Energy Market)
Dec 2025 - Jun 2026, Edinburgh

  • Pioneered a ~14% relative error hierarchical forecast for energy consumption/generation, shipped with a confidence score.
  • Delivered a tool for assessing PPA terms and advising clients on commercial risk, including Engie and Microsoft. Monte Carlo simulations, cannibalisation factors and various forecasts.

Data Science Manager - Scoring

CDP
Jan 2025 - Mar 2026, Edinburgh

  • Achieved 100x speedup, 30x automated test coverage, ~30% of my team’s annual capacity freed by redesigning the scoring engine. Moved from local testing to CI/CD on Azure DevOps.
  • Led a team of 4, automating scoring for ~30k accounts yearly, totalling an estimated $525M revenue per year. I brought my team’s code to CDP’s backend engineering standards.
  • Moved QA 3 months earlier in a 12 months release cycle with a synthetic answer generator covering ~500 of the hardest question variations per cycle.
  • Recovered a production failure carrying £20k/day of exposure in under 2h: diagnosed and time-boxed a fix against a management preference to roll back and re-scored 125 affected clients.
  • Allowed CDP to ship the Action Explorer, covering 1,000+ governments across 80+ countries by acting as GIS technical enabler. I made a case to the Director of Data, CPO and Legal Team that Overture Maps Foundation should be CDP’s geospatial source over Google Maps and won;

Senior Data Scientist - Product Lead (NLP & GIS)

Brainnwave
Jan 2023 - Jan 2025, Edinburgh

  • Automated 18 GIS indicators for an urban-planning methodology using infrastructure, real estate, population and satellite data for the City of Perth, and managed stakeholder presentations through to sale.
  • Halved the error on a NASA satellite product: fine-tuned MODIS44b.061 canopy coverage against official government methodology: RMSE reduced from 16% to 8%.
  • Classified 1M+ companies for Chicago, Hamilton (Ontario) and the University of Exeter by solving industry codes limitations using a custom LLM pipeline, pair-programmed with the CTO in two days and taken to commercial release in two months.

Senior Urban Data Scientist

The Business Of Cities
Oct 2021 - Jan 2023, London

  • Scaled TBoC’s branding report data coverage (total revenue £50k+) by 10x while cutting writing time by 80%. Processed 3.6M+ tweets and 20k+ news and search results across 10 cities using ML and automation. Clients include Oslo, Stockholm and Adelaide;
  • Informed 4000+ HSBC employees about worldwide covid recovery in 100 cities by automating the creation of a monthly analytics report up from 35 cities and days of manual work to under an hour;

Energy Data Scientist

Flexitricity
Feb 2020 - Mar 2021, Edinburgh

  • Our desk delivered higher revenue per MW than our closest direct competitor on 74% of days during my tenure, winning 17 of 18 months.
  • Delivered £3.6k/MW/year in gross revenue from gas power plants in the Balancing Mechanism alone via 10+ real-time price forecasting models running every 30 minutes, 24/7, to price ~550MW of flexible power across 4 markets: Day-Ahead, Balancing Mechanism, Capacity Market and Ancillary Services.
  • Produced a total of ~50 pieces of market insight and key performance numbers for the trading team, including revenue benchmarking against competitor units. I demonstrated that a bid-chasing strategy was net-negative and it was dropped, and showed our 49MW battery could be paid to charge in the BM rather than paying ~£23k in a single month.

Education

MSc, Physics of Complex Systems

École normale supérieure de Lyon

  • Class of 2018, Magna Cum Laude
  • ENS carries the same academic reputation as Oxford or Cambridge: entry by national competitive exam, ranked the world’s 5th best small university by Times Higher Education.
  • Master’s thesis on game theory and phase transitions: arXiv:1909.06858

MEng, Nanoscale engineering

Centrale Lyon

  • Class of 2016, Magna Cum Laude
  • Master’s thesis on superconducting nanowires for quantum optics

Skills

  • Machine Learning, Data Science, Data Mining, Data Analytics, Data Engineering, Deep Learning, NLP, LLMs, GIS, Forecasting, Time Series, Complex Systems;
  • MLOps, DevOps, CI/CD, System Design, Data Architecture, Data Governance;
  • Python, SQL, Bash, scikit-learn, PyTorch, Tensorflow, Keras, Transformers, Huggingface, Geopandas, Osmnx, H3, Overture, Docker, Airflow, MLflow, Github, AWS, Azure, GCP, Excel;
  • Energy, Urban Planning and Policy, Transportation, Industry Classification, ESG reporting, Workshop Design, Product Design, Facilitation;