Learning Path

Explore computational design research with the open-source tools developed by the Design Research Collective. Start with a runnable tutorial, adapt an example, and follow the parts of the ecosystem that match your interests.

Start with a tutorial. Extend an example. Share an improvement if it is useful to others.

Choose a tutorial Set up VS Code Workshop setup and preflight

Four Steps

1. Run One Example

Install the published design-research package and confirm that you can load a packaged problem. The VS Code setup guide provides the most direct first-user path. Python 3.12 or newer is required.

If you already work comfortably from a terminal, use the Quickstart instead.

2. Choose A Tutorial

All tutorials include runnable source material and displayed results. Most run offline. The local-model tutorial is clearly marked and can be skipped.

Track	Start here	What you will practice	Runtime
Design problems	Map packaged text problems	Loading a research corpus, representing text, and interpreting a map	Offline
Design grammars	Explore a truss grammar	Applying rules and inspecting a structured design state	Offline
Agent workflows	Build a deterministic workflow	Composing and tracing a small reasoning process	Offline
Experiments	Test a Monty Hall strategy	Conditions, replication, seeding, and reproducible results	Offline
Analysis	Measure coder reliability	Preparing observations and quantifying agreement	Offline
Complete workflow	Run a full-stack study	Connecting problems, agents, experiments, and analysis	Offline
Local AI models	Try propose/critic	Running and comparing a model-backed agent pattern	Ollama

The complete tutorial index includes additional process-comparison and factorial-analysis paths.

3. Make The Example Yours

Once an example runs unchanged, make one deliberate extension. For example:

• select a different packaged problem;

• change an experimental condition or number of replications;

• add a small deterministic agent step;

• compare another metric or visualization; or

• improve an explanation that was difficult to follow.

Record what you changed, what you expected, and what happened. A working extension and a short explanation are a complete learning outcome.

4. Share An Improvement

If your extension could help other users, share it through the usual open-source workflow:

1. Read the contribution guide.

2. Keep the change focused and explain the user-facing improvement.

3. Run the checks requested by the repository you changed.

4. Open a pull request through the repository’s normal GitHub workflow.

Contributions are reviewed for project fit and may be revised before they are accepted.

Where The Libraries Fit

The umbrella package connects four focused libraries:

• design-research-problems documentation (source) provides prompts, design tasks, grammars, benchmarks, and evaluators.

• design-research-agents documentation (source) provides AI participants, workflows, tools, and traceable reasoning patterns.

• design-research-experiments documentation (source) defines hypotheses, conditions, replications, execution, and study artifacts.

• design-research-analysis documentation (source) validates, transforms, models, and visualizes exported study data.

Begin with the umbrella tutorials. Move into a component repository only when you want to understand or change that layer in more detail.