Pipeline Core is up to 88% cheaper than industry standard deployments and runs at sub 25ms latency

Overview

Pipeline Core (PCore) is a machine learning compute platform designed for Data Scientists or ML/AI engineers to take their python code to production. By automating the creation, management, and post deployment lifecycle of ML models it reduces cloud costs and the dependency of a Data Science team on infrastructure engineers.

PCore’s current focus is to optimise ML inference workloads, but it does run any python code so is compatible with generic data science processing.

This whitepaper outlines the architecture, performance and use cases of PCore. It will be shown that PCore can reduce the cloud cost of industry standard deployments up to 88% and can operate at sub 25ms latency.