This repository is a curated collection of hands-on experiments and example outputs from a wide range of open-source AI frameworks that I have actively explored and implemented.
Rather than focusing on a single tool or library, this repo is designed to:
- Showcase practical usage patterns of modern AI frameworks
- Demonstrate how different tools can act as building blocks for real-world systems
- Serve as a growing knowledge base for agents, RAG pipelines, fine-tuning workflows, and applied AI techniques
All frameworks included here are open-source, making this repository both a learning resource and a collaborative space for experimentation.
I actively use my free time to:
- Learn and experiment with new AI frameworks and libraries
- Explore agentic systems, vibe-coding agents, retrieval-augmented generation, and model fine-tuning
- Combine individual tools into cohesive, real-world–oriented workflows
This repository reflects an ongoing effort to connect these individual pieces into practical, problem-solving pipelines, rather than isolated demos.
This is a learning-first, community-friendly repository.
Suggestions are always welcome, including:
- Code improvements or optimizations
- Frameworks or tools worth exploring
- Architectural or design feedback
Feel free to connect with me via LinkedIn or email to discuss ideas, improvements, or collaborations.
Below are examples of frameworks currently demonstrated in this repository. Each section focuses on how the framework is used, not a deep dive into its internals.
| Framework | Category | Problem Solved | AI Workflow Type | What Is Demonstrated in This Repo | Status |
|---|---|---|---|---|---|
| Docling | Document Processing | Extracting structured information from complex documents (PDFs, rich files) | Document AI · RAG Preparation · Data Ingestion | Document-to-text and document-to-structure pipelines producing AI-ready outputs | Active |
| Crawl4AI | Web Crawling | Reliable extraction of structured and unstructured web data | Web Data Ingestion · RAG · Agent Tooling | Website crawling and conversion of web content into clean Markdown for LLM pipelines | Active |
| (Future Framework) | (e.g., Agent Framework) | (Problem it addresses) | (Agent · RAG · Fine-tuning · Multimodal, etc.) | (What is implemented or tested) | Planned |
ℹ️ Notes
- The “Status” column indicates whether the framework is actively used in experiments or pipelines at present.
- Each framework has a dedicated section below with a short explanation and links to official documentation.
More frameworks and experiments will be added as this repository evolves.
📄 Docling Usage
This repository includes applied use cases of Docling, a document processing framework for parsing and converting diverse document formats into structured, AI-ready representations.
Here, Docling is used to demonstrate document-to-text and document-to-structure workflows, showing how complex documents (such as PDFs and other rich formats) can be transformed into outputs suitable for downstream tasks like analysis, retrieval, or integration with generative AI systems.
The emphasis is on practical usage and integration patterns, rather than an exhaustive exploration of Docling’s internal features.
🔗 Official Repository & Documentation https://github.com/docling-project/docling
🕷️ Crawl4AI Usage
This repository includes practical use cases of Crawl4AI, an open-source web crawling framework designed to convert websites into clean, LLM-ready Markdown.
Crawl4AI is used here as a web-to-data utility, demonstrating how structured and unstructured web content can be reliably collected and prepared for:
- RAG pipelines
- Agent-based systems
- Downstream AI and data processing workflows
The focus remains on usage patterns and integration, not framework internals.
🔗 Official Repository & Documentation https://github.com/unclecode/crawl4ai
- Framework-agnostic: No single tool dominates the repository
- Hands-on first: Emphasis on runnable examples and outputs
- Real-world mindset: Each experiment is treated as a building block, not a toy example
- Continuously evolving: New frameworks and techniques are added over time