Skip to content

README.md

Latest commit

 

History

History
140 lines (110 loc) · 4.67 KB

README.md

File metadata and controls

140 lines (110 loc) · 4.67 KB

06. Projects

Complete end-to-end projects demonstrating Ray's capabilities across different domains. These projects combine multiple Ray libraries and concepts to build production-ready applications.

Directory Structure

06-projects/
├── end-to-end-llms/                    # LLM fine-tuning and deployment
│   ├── 01_Finetuning_LLMs.ipynb        # Fine-tuning LLMs with Ray
│   ├── 02_Preparing_Data.ipynb         # Data preparation for LLMs
│   ├── 03_Evaluating_LLMs.ipynb        # Evaluating LLM performance
│   ├── 04_Deploying_LLMs.ipynb         # Deploying LLMs with Ray Serve
│   └── bonus/
│       └── MLOps_and_LLMs.ipynb        # MLOps practices for LLMs
├── rag/                                # Retrieval-Augmented Generation
│   ├── 01_Intro_to_RAG.ipynb           # Introduction to RAG
│   ├── 02_Index_Data.ipynb             # Indexing data for retrieval
│   ├── 03_Build_RAG.ipynb              # Building RAG pipelines
│   ├── 04_Deploy_RAG.ipynb             # Deploying RAG applications
│   └── bonus/
│       └── Deploy_LLM.ipynb            # Deploying LLMs
├── ray-tune/                           # Hyperparameter tuning
│   ├── Intro_Tune.ipynb                # Introduction to Ray Tune
│   └── Tune_Train.ipynb                # Integrating Tune with Train
├── video-highlight-generator/          # AI-powered video processing
│   └── README.md                       # Full project documentation
└── README.md

Projects Overview

1. End-to-End LLMs

Folder: end-to-end-llms/

A complete workflow for working with Large Language Models using Ray:

Notebook Description
01_Finetuning_LLMs.ipynb Fine-tune LLMs using Ray Train for distributed training
02_Preparing_Data.ipynb Prepare and preprocess data with Ray Data
03_Evaluating_LLMs.ipynb Evaluate LLM performance at scale
04_Deploying_LLMs.ipynb Deploy LLMs with Ray Serve
bonus/MLOps_and_LLMs.ipynb MLOps best practices for LLM workflows

Topics covered:

  • Distributed fine-tuning with Ray Train
  • Data preprocessing with Ray Data
  • Model evaluation and benchmarking
  • Production deployment with Ray Serve
  • MLOps practices and monitoring

2. RAG (Retrieval-Augmented Generation)

Folder: rag/

Build and deploy RAG applications that combine retrieval systems with LLMs:

Notebook Description
01_Intro_to_RAG.ipynb Introduction to RAG architecture
02_Index_Data.ipynb Build vector indices with Ray Data
03_Build_RAG.ipynb Create RAG pipelines
04_Deploy_RAG.ipynb Deploy RAG with Ray Serve
bonus/Deploy_LLM.ipynb Deploy the underlying LLM

Topics covered:

  • RAG architecture and components
  • Document chunking and embedding
  • Vector database integration
  • Query processing and retrieval
  • End-to-end RAG deployment

3. Ray Tune - Hyperparameter Tuning

Folder: ray-tune/

Learn to optimize model hyperparameters with Ray Tune:

Notebook Description
Intro_Tune.ipynb Introduction to Ray Tune
Tune_Train.ipynb Integrating Tune with Ray Train

Topics covered:

  • Hyperparameter search algorithms
  • Search space definitions
  • Early stopping strategies
  • Integration with Ray Train
  • Experiment tracking

4. Video Highlight Generator

Folder: video-highlight-generator/

An AI-powered system that automatically creates highlight reels from videos:

Key Features:

  • Ray Actors for distributed ML inference
  • MobileNetV3 for feature extraction
  • Multi-signal highlight detection
  • YouTube video support
  • Cluster compatibility

→ Full Documentation

Prerequisites

  • Python 3.12+ installed
  • Ray (latest version) installed via uv (see main README)
  • Understanding of Ray Core, Train, Data, and Serve concepts
  • Project-specific dependencies (see individual READMEs)

Getting Started

Each project can be run independently:

# End-to-End LLMs
cd 06-projects/end-to-end-llms
jupyter notebook

# RAG Project
cd 06-projects/rag
jupyter notebook

# Ray Tune
cd 06-projects/ray-tune
jupyter notebook

# Video Highlight Generator
cd 06-projects/video-highlight-generator
pip install -r requirements.txt
python demo.py

Resources