ado
ibm.github.io
- API
- Self-hosted
What ado is
ado is a free, open-source Python platform for designing computational experiment campaigns and running them at scale. Its workflow covers selecting experiments, defining a parameter space, configuring a strategy, executing runs, and analysing results. Distributed researchers and engineers can collaborate and share experiment data. A plugin model lets users extend experiments and operators with Python functions. The sample store saves measurements and can reuse earlier results through memoization. ado supports validated schemas, dry runs, structured results, and provenance records describing resource relationships and plugin versions. Documentation covers remote execution on Ray clusters, deployment with Kubernetes or OpenShift, and a distributed MySQL backend. Bundled agent skills can help coding assistants formulate campaigns, run experiments, query data, and analyse findings. Listed applications include fine-tuning and inference benchmarking and predictive performance model creation.
Who it suits
ado suits researchers and benchmarkers who explore parameter spaces, run computational experiments, and analyse results. It also fits distributed engineering teams needing extensibility and recorded provenance.
Good to know, and what to watch
- Free and open source
- Python plugin extensibility
- Memoized result reuse
- Provenance and validated schemas
- Ray, Kubernetes, and OpenShift support
- Requires Python-based workflows
- Documentation targets researchers and benchmarkers
- Backend setup includes distributed MySQL
The verdict
ado provides a structured experiment workflow with extensibility, reusable results, and provenance tracking. Its deployment and backend documentation support teams operating experiments across distributed infrastructure.
What the maker publishes
- Agent workflows
- Bundled agent skills can help AI coding assistants formulate research campaigns, run experiments, query data, and analyse results. source
- Data backend
- The backend guide covers a distributed MySQL setup and onboarding projects to it. source
- Examples
- Listed use cases include fine-tuning and inference performance benchmarking and predictive performance model creation. source
- Extensibility
- Experiments and operators can be extended through a plugin model using Python functions. source
- Integrations
- The documentation describes remote execution on Ray clusters and backend deployment with Kubernetes or OpenShift. source
- Intended users
- The user guide is aimed at researchers and benchmarkers who want to run experiments, explore parameter spaces, and analyse results. source
- Purpose
- ado is a Python platform for designing computational experiment campaigns and executing them at scale. source
- Research workflow
- Its documented workflow covers choosing experiments, defining a parameter space, configuring a strategy, executing experiments, and analysing results. source
- Result storage
- The sample store saves measurement results and can reuse existing results through memoization. source
- Support and contributions
- The project welcomes contributions and directs users to its GitHub issue tracker for feature proposals and bug tracking. source
- Team collaboration
- ado enables distributed teams of researchers and engineers to collaborate, run experiments, and share data. source
- Validation and provenance
- The documentation says ado supports validated schemas, dry runs, structured results, and provenance that records resource relationships and plugin versions. source
Checked against the maker's pages on 4 Oct 2026.