01
Create a new workflow
Import from a YAML file, copy a previous workflow, or start from scratch with a built-in template.
CIQ trusted by:
Run your most demanding workloads, anywhere.
Engineers, data scientists, and researchers move from problem to result faster through a web UI without ever touching infrastructure. Administrators see and control every environment from one control plane.
Build and ship from day one
Anyone on your team submits jobs, manages workflows, and accesses results through a web UI with no prerequisite infrastructure expertise.
Data follows your jobs
Data moves to wherever the job runs automatically so pipelines arrive fully provisioned with everything they need.
GPUs run at capacity, not idle
GPU resources allocate across training and inference workloads automatically to maximize utilization across every environment.
Full AI lifecycle in one workflow
Train a model, expose it as a live inference service, and access outputs in a notebook, all from a single workflow definition.
Start running in minutes
Validated templates for AI/ML, HPC, and interactive analysis get your first pipeline running immediately.
Multi-node at any scale
Coordinate distributed workloads across heterogeneous clusters without managing the underlying topology.
Secrets scoped to your jobs
Manage secrets directly in Fuzzball, scoped to jobs and environments, with no ancillary infrastructure required.
Deploy across every cloud
Run workloads across CoreWeave, AWS, and OCI from one control plane, routed automatically to the optimal destination.
Cloud scale. On-premises control.
Fuzzball Federate routes each job to the optimal environment based on cost, performance, and data locality across on-premises clusters and multi-cloud infrastructure. Develop in the cloud, deploy on-premises for production, and burst to cloud on demand. The workflow definition stays the same across all environments.
Intelligent job routing
Fuzzball evaluates every available environment and automatically routes each job to its optimal destination.
On-premises sovereignty
Placement policies enforce where regulated workloads run so models and data stay on infrastructure you control.
One workflow, every environment
The same workflow definition runs on every cluster, every cloud, and in every region you operate.
Fuzzball in action: Build and deploy your workflows with ease
Fuzzball's web UI is designed for usability and standardization. Anyone can build, manage, and reuse workflows with just a few clicks:
02
Define your jobs
Parameterize and customize every step of your workflow to match the needs of your jobs.
03
Customize your environment
Specify infrastructure requirements, from instance sizing to storage volumes, ensuring your workflow runs efficiently.
04
Save reusable workflow templates
Workflows can be made into application templates, enabling applications and workflows to be used by anyone.
05
Run, share, and reuse
Fuzzball workflows are modular and portable. Run them on any Fuzzball cluster, chain outputs as inputs to the next workflow, and share templates across your team.
Orchestrate any workload, on any infrastructure.
AI training and inference
Define training and inference in one workflow. When training completes, Fuzzball provisions a live inference endpoint automatically. Access results through Jupyter notebooks or virtual desktops directly in the browser, with your models on infrastructure you control throughout.
Traditional HPC alongside AI
Run molecular dynamics simulations alongside machine learning training pipelines on the same cluster in shared environments.
Bioinformatics pipelines
Run Nextflow workflows with automatic resource scaling, data management, and reproducible environments from the Workflow Catalog.
Interactive analysis with existing schedulers
Submit batch jobs through PBS or Slurm, your existing scheduler stays as the backend provisioner. Access outputs through Jupyter notebooks served directly through Fuzzball. Modern AI orchestration layers on top of the infrastructure your team already manages.
Hybrid cloud bursting
Develop locally, scale to cloud when capacity demands it, and return to on-premises for production or compliance requirements. The same workflow runs across every environment.
Build on a platform built for scale.
Works with your applications
Pre-built integrations and workflow templates get your team started on familiar tools without custom configuration.
Works with your pipelines
API-driven automation lets you submit and manage jobs programmatically from existing CI/CD tooling.
Works with your infrastructure
Built-in support for PBS Professional, Slurm, AWS, S3, Docker, Apptainer, Nextflow, and more. No custom drivers or connectors required.
FYR Diagnostics: 100x the throughput for production-scale genomics
FYR's EV-Omics platform processes complex multi-omics data to identify precision medicine targets. With Fuzzball, their team runs reproducible analytical pipelines at production scale, cutting cycle times from days to hours.
What's new in Fuzzball
Multi-cloud deployment
Fuzzball now deploys across CoreWeave, AWS, and OCI from a single control plane. Run any workload on any cloud, routed automatically to its optimal destination, with the same workflow definition across every environment.
Service Endpoints
Train a model and deploy it as a live inference endpoint from one workflow. Access results in the browser with your models on infrastructure you control.
PBS/Slurm Integration
Fuzzball integrates with PBS and Slurm as backend provisioners so your team gets modern AI orchestration without replacing the scheduler managing your existing resources.
Workflow Catalog
Pre-built templates for HPC, AI/ML, and interactive environments get common workloads running in minutes.
Fuzzball resources
Tech session
A 45-minute walkthrough of Fuzzball's core capabilities, federation architecture, and the full AI workload lifecycle.
Download the Fuzzball guide.
Download the Fuzzball guide and learn more about Fuzzball!
