William Smith
Senior Developer at Roscommon Analytics LLC
London, England, United Kingdom
Summary
👤
Senior🎓
Top SchoolWilliam Smith is a Senior Developer based in London with 14 years of experience building high-performance systems for trading and risk across power, gas and commodities markets. He combines a strong quantitative background (PhD in Mathematical Finance) with hands-on engineering—designing scalable data lakes, yield curve calibration engines, and a 1TB+ ERA5 weather datastore powering multi-desk analytics. At Roscommon Analytics he led AI policy and rollout of coding assistants and third-party LLMs while extending gas and hydro models to handle hundreds of scenarios with no performance loss. Former head of development for FIS Investment Risk, he has deep expertise in cross-asset market data architectures, VaR, and production ML pipelines. An active open-source contributor, he has improved low-level Linux block caching behavior in Facebook’s flashcache and refined datetime feature engineering in AutoGluon, reflecting a comfort across systems and ML stacks. Pragmatic and curious, he blends academic rigour with production-grade engineering to turn complex quantitative problems into reliable, auditable software.
14 years of coding experience
18 years of employment as a software developer
MA Computer Science, MA Computer Science at University of OxfordPhD Mathematical Finance, PhD Mathematical Finance at Birkbeck, University of LondonBSc Computer Science, BSc Computer Science at University of Warwick
English, Spanish, Malay, German, Dutch, Afrikaans
Github Skills (19)
python10
pandas10
machine-learning10
system-design10
feature-engineering10
c1110
linux-kernel10
c1710
automated-machine-learning10
testing9
debug9
debugging9
pytest9
scikit9
scikit-learn9
Programming languages (3)
Github contributions (5)
facebookarchive/flashcache
Sep 2011 - Oct 2011
A general purpose, write-back block cache for Linux.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:11 commits in 1 month
Contributions summary:William contributed to the `flashcache` repository, a write-back block cache for Linux. Their work primarily involved modifying and enhancing the core functionality of the cache. This included implementing features like skipping sequential I/O, refining metadata handling, and improving statistical reporting and debugging output. The user also addressed code duplication and incorporated sysctl configurations.
linuxcache-blockpurposewrite-backcache
autogluon/autogluon
Dec 2021 - Mar 2022
Fast and Accurate ML in 3 Lines of Code
Role in this project:ML Engineer
ML EngineerContributions:13 reviews, 7 commits, 11 PRs in 3 months
Contributions summary:William primarily focused on improving the datetime feature handling within the AutoGluon library, specifically within the context of feature engineering for machine learning tasks. This involved modifying code in several test files (test_datetime.py, test_auto_ml_pipeline.py, test_bulk.py, and test_pipeline.py) and the core datetime feature generator itself. The user also contributed to the documentation by adding basic feature engineering tutorials and corrected minor typos. In addition, the user addressed a deprecation warning and capped dependency versions.
forecastingimage-textmlppythonmeta-learning
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.