Adrian Letchford

About me

I'm a quantitative developer and data scientist with over a decade of experience solving hard problems with technology. My career has taken me from academia, to technology startups, and into finance - giving me a rare mix of research depth, product focus, and market experience.

I've worked across roles spanning data science, full-stack engineering, data engineering, and leadership, which means I can take ideas from first principles through to production systems.

Whether I'm building models, engineering data pipelines, or writing about quant topics, I care about doing work that's both technically strong and human-centred.

Active projects

Open Source Quant

I write about quantitative finance, data, and engineering at osquant.com, where ~1,000 readers each month explore my articles, tools, and research. Inspired by my academic background, it's driven by a belief that rigorous, useful knowledge should be accessible without unnecessary gatekeeping.

streetview

I maintain an unofficial Google Street View API Python client that lets you download both current and historical panoramas — something the official API doesn't support. Built for researchers, developers, and anyone who needs more than the standard Street View offering. Available on PyPI and GitHub.

recx

I created a Python library to make my own data pipelines more robust. The library "recx" performs reconciliation tests on DataFrames. This means you can be sure that refreshes to your data hasn't created unknown changes to the data. Available on PyPI and GitHub.

Selected works

software ibportal (2022) A Dockerized version of Interactive Brokers' REST API Web Gateway.

video Airtasker: Meet "Carl!" The AI Classifier! 2018.
I talked to Amazon Web Services about Carl, one of the awesome machine learning tools that we built at Airtasker. Carl is an algorithm that classifies the tasks posted to the Airatasker platform - named after Carl Linnaeus.

website CoventryInData.com 2016. A series of data visualisations illustrating the City of Coventry.

paper Adrian Letchford, Tobias Preis and Helen Susannah Moat. 2016 Quantifying the Search Behaviour of Different Demographics Using Google Correlate. PLoS ONE 11 (2): e0149025.

paper Adrian Letchford, Tobias Preis and Helen Susanna Moat. 2016. The advantage of simple paper abstracts. Journal of Informetrics 2 (1). pp. 1-8.

course Robot Races. 2015. I taught a crash course on programming and robotics with Lego robots. Classes were held at the University of Warwick and aimed for school years 6+. At the end, the kids race their robots against the clock through an obstical course using a combination of light and ultrasonic sensors.

show Claire Rocks and Adrian Letchford. The University of Warwick 2015 Christmas Lecture on Robotics.
I co-hosted a 2015 Christmas Lecture at the University of Warwick. The show was about robotics, and all the different parts that make a functioning robot, as well as the roles these robots now play in society.

paper Christian M. Alis, Adrian Letchford, Helen Susannah Moat and Tobias Preis. 2015. Estimating tourism statistics with Wikipedia page views. Proceedings of the ACM Web Science Conference Article No.: 33. pp. 1-2.

paper Adrian Letchford, Helen Susanna Moat and Tobias Preis. 2015. The advantage of short paper titles. Royal Society Open Science 10: 2150266.

The paper was written about by:

paper Adrian Letchford, Junbin Gao, and Lihong Zheng. 2014. Smoothing Security Prices. 22nd International Conference on Pattern Recognition. pp. 1037-1042.

paper Adrian Letchford, Junbin Gao, and Lihong Zheng. 2013. Filtering financial time series by least squares. Int. J. Mach. Learn. & Cyber. 4. pp. 149-154.

paper Adrian Letchford, Junbin Gao and Lihong Zheng. 2012. Optimizing the moving averages. The 2012 International Joint Conference on Neural Networks (IJCNN).

paper Adrian Letchford, Junbin Gao and Lihong Zheng. 2011. Penalized least squares for smoothing financial time series. AI 2011: Advances in Artificial Intelligence. 7106.