Forecasting time-series with Arrow and WASM

This session is a presentation of novel concept of time-series processing pipeline for MLOps

– Apache arrow + s3 in backed to process large number of large files

– polars or duckdb for data manipulation

– super fast task runner written in Rust

– WASM tasks in which we do forecasting Focus on current state of the art of Python in WASM

– possible performance improvement in production deployments of time-series models

Technical Level: Introductory level/students (some technical knowledge needed)

Supported by