{"id":19760,"date":"2024-12-24T15:39:52","date_gmt":"2024-12-24T10:09:52","guid":{"rendered":"https:\/\/opstree.com\/blog\/?p=19760"},"modified":"2025-08-14T23:04:25","modified_gmt":"2025-08-14T17:34:25","slug":"end-to-end-rag-solution-with-aws-bedrock-and-langchai","status":"publish","type":"post","link":"https:\/\/opstree.com\/blog\/2024\/12\/24\/end-to-end-rag-solution-with-aws-bedrock-and-langchai\/","title":{"rendered":"End-to-End RAG Solution with AWS Bedrock and LangChain"},"content":{"rendered":"
\n
\n

Introduction<\/b><\/h2>\n

In this blog, we\u2019ll explore the powerful concept of Retrieval-Augmented Generation (RAG) and how it enhances the capabilities of large language models by integrating real-time, external knowledge sources. You’ll also learn how to build an end-to-end application that leverages this approach for practical use.<\/span>\u00a0<\/span><\/p>\n

We\u2019ll begin by understanding what RAG is, how it works, and why it’s gaining popularity for building more accurate and context-aware AI solutions<\/strong><\/a>. RAG combines the strengths of information retrieval and text generation, enabling language models to reference external, up-to-date knowledge bases beyond their original training data, making outputs more reliable and factually accurate.<\/span>\u00a0<\/span><\/p>\n

As a practical demonstration, we\u2019ll walk through building a custom RAG application that can intelligently query information from your own PDF documents. To achieve this, we’ll use the AWS Bedrock Llama 3 8B Instruct model, along with the LangChain framework and Streamlit for a user-friendly interface.<\/span>\u00a0<\/span><\/p>\n<\/div>\n

Key Technologies For End-to-End RAG Solution<\/h2>\n
\n

1.<\/span><\/strong><\/span> Streamlit<\/span>:<\/span><\/span><\/strong>
a.<\/span><\/span> Interactive front<\/span>–<\/span>end for the application.<\/span><\/span>
b.<\/span><\/span> Simple yet<\/span> powerful framework for building Python web<\/span>
apps.<\/span><\/span>
2.<\/span><\/strong><\/span> LangChain<\/span>:<\/span><\/span><\/strong>
a.<\/span><\/span> Framework for creating LLM<\/span>–<\/span>powered workflows.<\/span><\/span>
b.<\/span><\/span> Provides seamless integration with AWS Bedrock<\/strong><\/a>.<\/span><\/span>
3.<\/span><\/strong><\/span> AWS Bedrock<\/span>:<\/span><\/span><\/strong>
a.<\/span><\/span> State<\/span>–<\/span>of<\/span>–<\/span>the<\/span>–<\/span>art LLM platform.<\/span><\/span>
b.<\/span><\/span> Powered by the highly efficient Llama 3 8B<\/span> Instruct model.<\/span><\/span><\/p>\n

<\/div>\n<\/div>\n
Let\u2019s get started! Implementing this application involves three key components, each designed to streamline setup and ensure best practices. With the right AWS consulting service<\/strong><\/a>, you can efficiently plan, deploy, and optimize each component for a secure and scalable solution.”<\/div>\n<\/div>\n

<\/p>\n

\n

1. Create a vector store<\/span><\/span><\/h3>\n
\"\"
<\/span>
Load<\/span>–<\/span>>Transform<\/span>–<\/span>>Embed.<\/span><\/span>
We will use the FAISS vector database,<\/span> t<\/span>o<\/span> efficiently handle queries,<\/span>
the text is tokenized and embedded into a vector store using FAISS<\/span>
(Facebook AI Similarity Search).<\/span><\/span><\/div>\n<\/div>\n
<\/div>\n
[ Find More About: Modernizing Legacy Systems on AWS<\/a>]<\/strong><\/div>\n
\n


2. Query vector store \u201cRetrieve most similar\u201d<\/span><\/span>
<\/span><\/h3>\n

\"\"<\/p>\n

The way to handle this at query time, embed the unstructured<\/span>
query and retrieve the embedding vector that is most similar to <\/span><\/span>the embedded query. A vector stores embed data and performs a<\/div>\n<\/div>\n
\n
vector search for you.<\/span><\/span><\/div>\n

<\/span>3.<\/span> Response generation using LLM:<\/span><\/span><\/h3>\n

 <\/p>\n

\"\"<\/div>\n

Imports and Setup<\/span><\/span>\u00a0<\/span><\/h3>\n
\"\"<\/div>\n
<\/div>\n<\/div>\n
\n
    \n
  1. os<\/span><\/strong>:<\/strong> Used for handling file paths and checking if files exist on disk.<\/span><\/span>
    <\/span><\/li>\n
  2. pickle<\/span><\/strong>:<\/strong> A Python library for serializing and deserializing Python objects to <\/span>store\/retrieve embeddings.<\/span><\/span>
    <\/span><\/li>\n
  3. boto3<\/span><\/strong>:<\/strong> AWS SDK for Python; used to interact with Amazon Bedrock services.<\/span><\/span>
    <\/span><\/li>\n
  4. streamlit<\/span><\/strong>:<\/strong> A library for creating web apps for data science and machine learning <\/span>projects.<\/span><\/span>
    <\/span><\/li>\n
  5. Bedrock<\/span><\/strong>:<\/strong> Used to interact with Amazon Bedrock for deploying large language <\/a><\/span>models (LLMs).<\/span><\/span>
    <\/span><\/li>\n
  6. Bedrock Embeddings<\/span><\/strong>:<\/strong> To generate embeddings using Bedrock models.<\/span><\/span>
    <\/span><\/li>\n
  7. FAISS<\/span><\/strong>:<\/strong> A library for efficient similarity search and clustering of dense vectors.<\/span><\/span>
    <\/span><\/li>\n
  8. Recursive Character Text Splitter<\/span><\/strong>:<\/strong> Splits large text into manageable chunks for <\/span>embedding generation.<\/span><\/span>
    <\/span><\/li>\n
  9. Pdf Reader<\/span><\/strong>:<\/strong> From PyPDF2, used to extract text from PDF files.<\/span><\/span>
    <\/span><\/li>\n
  10. Prompt Template<\/span><\/strong>:<\/strong> Defines the structure of the prompt for the LLM.<\/span><\/span>
    <\/span><\/li>\n
  11. Retrieval QA<\/span><\/strong>:<\/strong> Combines a retriever and a chain to create a question<\/span>–<\/span>answering <\/span>system.<\/span><\/span>
    <\/span><\/li>\n
  12. Stuff Documents Chain<\/span><\/strong>:<\/strong> Combines multiple documents into a single context for <\/span>answering questions.<\/span><\/span>
    <\/span><\/li>\n
  13. LLM Chain<\/span><\/strong>:<\/strong> A chain that interacts with a language model using a defined prompt.<\/span><\/span><\/li>\n
  14. Initialize Bedrock and Embedding Models<\/strong><\/strong>\"\"<\/li>\n
  15. Initializes an Amazon Bedrock<\/a> client using boto3 to interact with Bedrock services.<\/span><\/span><\/li>\n
  16. Initializes the bedrock titan embedding model <\/span>amazon.titan<\/span>–<\/span>embed<\/span>–<\/span>text<\/span>–<\/span>v1<\/span> for <\/span>generating vector embeddings of text.
    \n<\/span><\/span>
    \n\"\"<\/span><\/span><\/li>\n<\/ol>\n<\/div>\n
    \n

    Define Bedrock LLM and Prompt<\/span> template<\/span><\/span><\/h2>\n
    \n

    The application begins by initializing the <\/span>Amazon Bedrock Large Language Model (LLM)<\/span><\/b><\/a> using a specified <\/span>model_id<\/span>. A <\/span>custom prompt template<\/span><\/b> is also defined to guide the AI assistant on how to generate responses effectively.<\/span>\u00a0<\/span><\/p>\n

    This template outlines specific formatting instructions, such as:<\/span>\u00a0<\/span><\/p>\n