Felipe Angelim
Staff Data Scientist
Rio de Janeiro, Brazil
Summary
🤩
Rockstar🎓
Top SchoolFelipe Angelim is a Staff Data Scientist based in Rio de Janeiro with nine years of experience building production-grade forecasting and MLOps systems for finance and e-commerce. He combines strong mathematical foundations with cloud-native engineering, having architected Terraform/Kubernetes/Helm-based MLOps and ETL pipelines for a hedge fund and production forecasting at Mercado Libre. His work spans hierarchical time-series forecasting, Bayesian TAM estimation, multi-armed bandits for UX optimization, and even 3D object classification, with tangible impact such as a real estate strategy that raised R$315M and client-acquisition forecasts months in advance. An active core developer on sktime and creator of Prophetverse, he contributes to the most-used Python time-series library, improving core forecasters and hierarchical tooling. Collected from engineering training in France and Brazil, Felipe blends hands-on model development with platform-building and mentorship, favoring interpretable, deployable solutions that drive business decisions.
9 years of coding experience
7 years of employment as a software developer
Bachelor's degree Mechanical Engineering, Bachelor's degree Mechanical Engineering at Federal University of Rio de JaneiroMaster's Degree (Diplôme d'ingénieur) Ingénierie, Master's Degree (Diplôme d'ingénieur) Ingénierie at Centrale Lyon
Portuguese, English, French, Spanish
Github Skills (14)
scikit-learn10
pandas10
forecasting10
debug10
machine-learning10
time-series10
forecast10
python10
scikit10
data-science9
testing9
prophet9
data-mining8
github8
Programming languages (4)
Github contributions (5)
sktime/sktime
Jul 2021 - Apr 2025
A unified framework for machine learning with time series
Role in this project:
ML Engineer
ML EngineerContributions:33 reviews, 20 PRs, 7 pushes in 3 years 9 months
Contributions summary:Felipe primarily contributes to the `sktime` repository, a framework for time series machine learning. Their work involves fixing bugs in core forecasting components, such as the `_BaseWindowForecaster`, to handle in and out-of-sample predictions correctly. They also implement new estimators, like a placeholder for `HierarchicalProphet` in the `prophetverse` module, showing engagement with expanding the framework's forecasting capabilities and addressing various code issues. The user further contributes by creating dataset objects and addressing issues related to hierarchical data structures within existing transformers, demonstrating a solid grasp of the project's architecture and how to improve its functionality.
forecastingtime-series-analysistime-series-regressiondata-sciencedeep-learning
felipeangelimvieira/sktime
Jul 2023 - Mar 2025
A unified framework for machine learning with time series
Contributions:86 pushes, 15 branches in 1 year 9 months
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.