Thomas Boquet
Head Of Real Time Bidding at Trellis
Greater Montreal Metropolitan Area Canada
Summary
🤩
Rockstar🎓
Top SchoolThomas Boquet is a machine learning leader with 11 years of experience building production-grade ML systems, currently heading Real Time Bidding at Trellis. He blends applied research and full-stack ML engineering—shipping anomaly detection, time-series, and LTV models into real-world products across gaming, adtech, and industrial domains. Previously a staff ML engineer at Mistplay and senior data scientist at Unity, he has a research background from Element AI and a PhD-level foundation in decision sciences. Thomas contributes to core deep learning tooling (notably enhancements to the widely used Keras library for training, visualization and custom losses), reflecting hands-on expertise with model training and observability. He designs minimalist, efficient MLOps architectures that emphasize reproducibility, low-latency serving, and pragmatic monitoring. Based in the Greater Montreal area, he combines academic rigor with product-driven delivery and a knack for turning research ideas into scalable pipelines.
11 years of coding experience
12 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Decision Sciences, Doctor of Philosophy (Ph.D.) Decision Sciences at Université de Montréal - HEC MontréalBachelor of Science (B.S.) Economics, Bachelor of Science (B.S.) Economics at Université de MontréalMaster of Science (M.Sc.) Business intelligence, Master of Science (M.Sc.) Business intelligence at HEC Montréal
French, English
Stackoverflow
164reputation
18kreached
3answers
2questions
Github Skills (15)
neural-network10
callback10
keras10
machine-learning10
deep-learning10
trainings10
tensorflow10
python10
modeling10
data-science9
vggnet6
face-recognition6
theano6
setuptools6
cython6
Programming languages (7)
Github contributions (5)
keras-team/keras
Nov 2015 - Oct 2016
Deep Learning for humans
Role in this project:
ML Engineer
ML EngineerContributions:16 commits, 14 PRs, 271 comments in 10 months
Contributions summary:Thomas contributed to the Keras library by implementing and improving features related to deep learning model training and visualization. They fixed issues in model validation, integrated TensorBoard for enhanced model monitoring, and added support for custom loss functions. Furthermore, the user added support for fit_generator and fixed minor issues.
pythondata-sciencedeep-learningneural-networksmachine-learning
tboquet/python-alp
Apr 2016 - Mar 2017
Contributions:1 release, 902 commits, 112 PRs in 11 months
machine-learning-platformpythonlearning-platformdata-sciencemachine-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.