Floability Overview

Floability deploys scientific workflows as portable units called backpacks. A backpack keeps a workflow together with the specifications needed to prepare its software, locate its data, and connect it to computing resources.

This part of the tutorial progresses from using an existing backpack to creating one and, finally, generating initial specifications from an observed notebook execution.

Floability sessions

Time	Session	What you will do
10 min	Floability overview	Understand the deployment problem and backpack model
15 min	Run your first backpack	Deploy an existing notebook workflow
10 min	Structure of a backpack	Examine workflow, software, data, and compute specifications
20 min	Create and run a backpack	Package a workflow and run the resulting backpack
15 min	Generate a backpack automatically	Use Audit to create initial specifications from a working notebook environment
5 min	Wrap-up and questions	Review the portability model and next steps

The backpack model

my-backpack/
├── workflow/       notebook, Python script, or shell entrypoint
├── software/       Conda environment specification
├── data/           data sources, profiles, and integrity information
└── compute/        worker and resource requirements

The specifications are explicit and reviewable. Floability uses them to stage data, prepare and cache software environments, create an isolated run instance, launch TaskVine workers, and start the workflow interactively or execute it without a browser.

The goal is not to hide every site difference. The goal is to keep the application's portable requirements with the workflow while allowing site-specific storage, scheduler, network, and policy settings to be supplied at deployment time.

Run your first backpack →

After the tutorial

The live schedule uses the matrix example. If you want a more advanced workflow afterward, continue with MobileNet batch inference or browse the complete collection in Floability Hub.