# AutoML Reproducibility Hub

Browser-first research and educational software by Faramarz Kowsari for reproducible machine-learning experiments.

Canonical links:
- Live application: https://faramarzkowsari.github.io/automl-reproducibility-hub/
- Repository: https://github.com/FaramarzKowsari/automl-reproducibility-hub
- Official guidebook: https://faramarzkowsari.github.io/automl-reproducibility-hub/guidebook/
- Guidebook PDF: https://faramarzkowsari.github.io/automl-reproducibility-hub/guidebook/inside-automl-reproducibility-hub.pdf
- ORCID: https://orcid.org/0000-0003-1692-0453

Core reproducibility fields:
- deterministic random seed
- versioned dataset and SHA-256 fingerprint
- explicit model and preprocessing parameters
- task-appropriate metrics
- runtime and package versions
- exportable experiment manifest and Python reproduction script

Modes:
- S: Static reference mode with precomputed, integrity-labeled fixtures
- F: Full browser execution with Pyodide, scikit-learn, pandas, NumPy, and DuckDB-WASM

Official guidebook:
- Title: Inside AutoML Reproducibility Hub
- Subtitle: A Visual Guide to Reproducible Browser-Based Machine Learning
- Format: 10-page A4 infographic PDF
- Author: Faramarz Kowsari