RAG Pipeline with Retrieval and Generation

Scenario

Deploy an embedding model in OpenSearch to power vector-based document retrieval, then deploy a fine-tuned LLM on Bailian as the generation endpoint — together forming a Retrieval-Augmented Generation (RAG) application that retrieves relevant context and produces grounded responses.

How the products combine

1. opensearch · opensearch-deploy-model — OpenSearch — Deploy embedding model for inference

See opensearch/opensearch-deploy-model.

2. bailian · bailian-deploy-model — — Deploy custom or fine-tuned AI models as endpoints

See bailian/bailian-deploy-model.

Typical questions

• build RAG pipeline
• deploy RAG system
• 搭建RAG应用
• vector search plus LLM
• embedding retrieval and generation
• 部署检索增强生成
• knowledge base QA system
• deploy embedding and LLM together

Q: How do I build and deploy a RAG system that combines vector search with an LLM? A: You build and deploy a RAG system by combining OpenSearch and Bailian to handle retrieval and generation respectively. First, you deploy an embedding model in OpenSearch to power vector-based document retrieval. Next, you deploy a fine-tuned LLM on Bailian as the generation endpoint to produce grounded responses from the retrieved context.