Improve the readme around the data centric approach based on embedding and update the core principles

[Yann-CV]

This pull request updates the README.md to broaden the scope of Goldener from a machine learning (ML) focus to a more general artificial intelligence (AI) context. It also clarifies the library's core principles, emphasizes its data-centric and modality-agnostic approach, and adds detail about its extensibility and standard dependencies.

Key documentation updates:

Expanded scope and clarified positioning

• Updated all references from "Machine Learning (ML)" to "Artificial Intelligence (AI)" to reflect a broader application of the library. [1] [2]
• Added explanation that Goldener's features are based on semantic representations using embeddings from pre-trained/foundational models.

Design principles and extensibility

• Rewrote the design principles section to highlight modality-agnosticism, customizability, and reliance on standard dependencies (e.g., PyTorch, numpy, scikit-learn) for compatibility and ease of integration.
• Clarified the iterative and incremental nature of AI pipelines and how Goldener supports progressive batch processing, multipurpose embeddings, distributed processing, and on-demand pipeline access.

Vision and research

• Added a reference to the Goldener open research repository and described the vision of unifying data-centric processes for all data and task types.

---

Summary by cubic

Rewrites the README to position Goldener as a data-centric library built on embeddings from pretrained models and usable across the full lifecycle of artificial intelligence pipelines. Updates the core principles to emphasize modality-agnosticism, customizability, and standard dependencies, adds a link to the open research repository, and refreshes the brand image.

^{Written for commit a7eae12. Summary will update on new commits.}

[Copilot]

🟡 Changes recommended

The updated “Key design principles” section introduces Markdown list formatting problems and a few typos/grammar inconsistencies that will render poorly and reduce documentation clarity.

Get a fresh assessment by requesting another Copilot review.

Pull request overview

This PR refreshes README.md to position Goldener as a broader data-centric AI library (vs. ML-only), grounding its capabilities in embedding-based semantic representations and updating the documented design principles to emphasize modality-agnosticism, customizability, and standard dependencies.

Changes:

• Reframes Goldener’s scope from “Machine Learning (ML)” to “Artificial Intelligence (AI)” throughout the overview.
• Adds an embedding/semantic-representation explanation as the foundation for features and a link to the open research repository.
• Rewrites the “Key design principles” section to better describe extensibility and operational constraints (iteration, scale, distributed execution).

File summaries

File	Description
README.md	Updates positioning, adds embedding-based framing, and rewrites/expands the design principles section.

Review details

Suppressed comments (3)

README.md:61

• Markdown list formatting: the second lines for these bullets aren’t indented, so they won’t render as part of the bullet items. Also, adding commas improves readability.

- **Customizable**: All features leverage specifiable tools. These tools are defined following standard API patterns
allowing anyone to implement their own tools to customize the behavior of the features.
- **Standard dependencies**: Goldener is built on standard dependencies (PyTorch, numpy, scikit-learn, etc.) to ensure
compatibility with the most common AI pipelines limiting the need for new dependencies.

README.md:72

• Markdown list formatting + spelling: the wrapped lines aren’t indented (so they won’t be part of the bullet items), and "Multipurposes" should be "Multipurpose".

- **Progressive batch processing**: Each task can be stopped and restarted on demand (or failure).
Already computed results are not recomputed.
- **Multipurposes embeddings**: The same embeddings are used for different tasks (selection, splitting, monitoring, etc.).
They are computed once and stored for later use. This allows to save time and computing resources.

README.md:75

• Markdown list formatting: this bullet’s continuation line isn’t indented, so it will render as a separate paragraph. Also "On demand" is typically hyphenated as "On-demand" when used adjectivally.

- **On demand access to pipelines**: All processing pipelines are serializable.
They are stored and available whenever a new request is made.

• Files reviewed: 1/2 changed files
• Comments generated: 5
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[cubic-dev-ai]

All reported issues were addressed across 2 files

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 1 file (changes from recent commits).

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>