Shrinking gigabyte sized scikit-learn models for deployment

Jonas Haag, Pavel Zwerschke

Type/Track Talk pydata-pydata-scientific-libraries-stack

We present an open source library to shrink pickled scikit-learn and lightgbm models. Together with a distroless Docker build we can reduce the deployment size of machine learning applications up to 3x.

Level Domain Expertise Novice Python Skill Level Intermediate

Jonas Haag

Affiliation: QuantCo

Jonas Haag is a Data Engineer at QuantCo. He has a background in computer science and more than 10 years of experience with Python, particularly in web backend development and data engineering. In the past, he has co-founded a fintech startup and a software consultancy.

visit the speaker at: Github • Homepage

Pavel Zwerschke

Affiliation: QuantCo

Pavel is a data engineer at QuantCo who is currently studying Mathematics and Computer Science at KIT.

visit the speaker at: Github • Homepage