Sebastian Rhode

Using Containers to deploy AI Models across our microscopy platform

How do you guarantee identical AI results from the cloud to a local desktop? Learn how containers solved this critical reproducibility challenge for scientific imaging.

Using Containers to deploy AI Models across our microscopy platform
#1about 4 minutes

AI-powered computer vision workflows in modern microscopy

Zeiss uses AI for various microscopy tasks like classification and instance segmentation to analyze biological samples.

#2about 2 minutes

The challenge of analyzing terabyte-scale microscopy data

Automated microscopy workflows can generate terabytes of data from a single experiment, requiring powerful AI for quantitative analysis like cell counting.

#3about 3 minutes

Key requirements for reproducible AI model deployment

Users need robust and reproducible AI models that deliver consistent results across different platforms without requiring IT expertise.

#4about 3 minutes

Moving from model artifacts to containerized deployments

The previous method of deploying only model files created synchronization issues, leading to the adoption of containers to package models with all their dependencies.

#5about 3 minutes

Why containers are the ideal solution for AI deployment

Containers solve key challenges by enabling GPU access on Windows via WSL2, decoupling dependencies for different AI tasks, and simplifying client software maintenance.

#6about 3 minutes

The new workflow for training and deploying models

The new process involves training models in the cloud, which produces a container as the final artifact that is then downloaded and run by the client software.

#7about 2 minutes

Demonstrating the business value of containerization

This container-based approach allows users to access new AI algorithms faster without client updates, convincing stakeholders and enabling independent development cycles.

#8about 4 minutes

Key learnings for adopting container technology

Adopting containers is successful when it solves a real business problem, starts with smaller prototype projects to de-risk, and leverages mature, standardized technology.

Related jobs
Jobs that call for the skills explored in this talk.

Featured Partners

From learning to earning

Jobs that call for the skills explored in this talk.