Tomas De Vasconcelos
Engineering Manager - Machine Learning at Canva
Amsterdam, North Holland, Netherlands
Summary
🤩
Rockstar🎓
Top SchoolTomas De Vasconcelos is an engineering manager and machine learning leader with a decade of experience turning complex data science experiments into production-grade, reliable systems. He has led forecasting and ML infrastructure at Nike and now scales AI-driven content moderation at Canva, blending solutions architecture, MLOps, and software engineering best practices to speed iteration without sacrificing reliability. Comfortable across time-series forecasting, recommender systems, anomaly detection and Bayesian methods, he champions robust evaluation, idempotent pipelines, and clear data provenance to build stakeholder trust. An advocate for developer experience, he routinely shapes repository structure, CI/CD, and orchestration (Airflow, Kubernetes, SageMaker) to make teams more productive. Tomas is also an active open-source contributor to sktime, where he fixed stability issues and extended Prophet integration—reflecting a practical focus on tooling that bridges research and production. Based in Amsterdam and grounded in a strong mathematical background, he combines curiosity (a Feynman-inspired motto) with a pragmatic approach to delivering business impact.
10 years of coding experience
8 years of employment as a software developer
BSc Physics, BSc Physics at Royal Holloway, University of LondonHigh School - AFS Intercultural Program, High School - AFS Intercultural Program at River Ridge High School, IL, USAScience & Technology High School Diploma Science & Technology, Science & Technology High School Diploma Science & Technology at Salesianos de Manique, Lisbon, PortugalGraduate Mathematics, Graduate Mathematics at King's College London
English, Portuguese
Stackoverflow
679reputation
218kreached
17answers
0questions
Github Skills (17)
debug10
python10
scikit10
pandas10
machine-learning10
time-series10
forecasting10
sktime10
scikit-learn10
forecast10
pip6
game-physics6
pygame6
sql-server6
windows6
Programming languages (14)
Github contributions (5)
sktime/sktime
Sep 2020 - Oct 2024
A unified framework for machine learning with time series
Role in this project:
Data Scientist
Data ScientistContributions:20 reviews, 6 PRs, 65 comments in 4 years 1 month
Contributions summary:Tomas primarily contributed to bug fixes and enhancements within the sktime framework. They addressed a `FutureWarning` related to pandas upcasting, improving stability by resolving potential crashes and slow-downs in IDEs. The user also added the `fit_kwargs` parameter to the Prophet forecaster, enabling customization of the fitting process, and adjusted the documentation of `FunctionTransformer`. Additionally, they corrected tag handling in `IgnoreX` and addressed a potential error in `pd.infer_freq`.
forecastingtime-series-analysistime-series-regressiondata-sciencedeep-learning
tpvasconcelos/sparkypandy
Jun 2021 - Jul 2022
It's not spark, it's now pandas, it's just awkward...
Contributions:50 commits, 4 PRs, 44 pushes in 1 year
data-sciencepythonpandasspark
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.