Add comprehensive documentation and 24-session learning curriculum

[Kandil7]

Summary

Adds a complete documentation set and a 24-session learning curriculum for the LLM Engineer's Handbook, written against the actual source code in this repository.

What's included

• Curriculum and guides: docs/CURRICULUM.md, docs/GETTING_STARTED.md, and docs/README.md (documentation index).
• Code intelligence: docs/CODEBASE-INTELLIGENCE.md with architecture, graphs, and tooling references.
• 24 session walkthroughs under docs/sessions/, covering:
  • Foundations: project overview, domain layer (ODM), infrastructure layer (MongoDB/Qdrant).
  • Data engineering: web crawling, text preprocessing, feature engineering.
  • Datasets: instruction dataset generation, preference dataset for DPO.
  • RAG: advanced RAG architecture, embedding models and cross-encoders.
  • Training: SFT/QLoRA, DPO, AWS SageMaker deployment.
  • Inference: FastAPI service, RAG inference flow.
  • Monitoring: Comet ML experiment tracking, Opik tracing, model evaluation.
  • Production: Docker, CI/CD with GitHub Actions, ZenML orchestration.
  • Advanced: data warehouse backup/restore, performance tuning, security hardening.
• .gitignore: ignore the local graphify-out/ artifacts directory.

Structure

Each session follows the same format: learning objectives, architecture overview, per-file deep dives with code, hands-on steps, an exercise, a knowledge check, and next steps.

Notes

• This is a documentation-only change; no runtime code is modified.
• Content is grounded in the real modules, class names, and configuration values in the repository.