RAG in Practice: Working Example to Get You Started

2026-05-31 · Source: LLM on Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, medium

Summary

A practical, working RAG (Retrieval Augmented Generation) sample application is presented, implemented in Node.js with TypeScript, Next.js 16, and React 19. This Dockerized project utilizes OpenAI's gpt-4o-mini for the LLM and text-embedding-3-small (1536 dims) for embeddings, with Qdrant serving as the vector database using cosine similarity. It allows users to upload PDF, Markdown, or plain text files, then ask questions, with answers derived exclusively from the uploaded content. The application details its architecture, including API endpoints for querying, uploading, and listing documents, alongside UI components for file upload and query interaction. Key code snippets illustrate PDF parsing with LangChain's PDFLoader, document chunking, embedding generation, and similarity search.

Key takeaway

For AI Engineers or Software Engineers building RAG applications, especially with Node.js and TypeScript, this project offers a robust starting point. You can clone the GitHub repository, set up your `.env` file, and run it with `docker compose up` to quickly get a functional RAG pipeline. Review the provided code, particularly the `ingest.service.ts` and `rag.service.ts` files, to understand the core logic and adapt it for your specific use cases, addressing any TODOs for production readiness.

Key insights

A practical RAG pipeline example is provided, demonstrating a full-stack implementation with modern web technologies.

Principles

• RAG uses embeddings to find relevant document chunks for LLM context.
• Chunking documents (e.g., 400-500 words) is crucial for LLM input management.
• Similarity scoring filters weak matches, preventing LLM hallucinations.

Method

The RAG pipeline involves retrieving text, chunking documents, generating embeddings, storing them in a vector DB, then embedding user questions to search for relevant chunks, which are finally fed to an LLM for answer generation.

In practice

• Use `PDFLoader` from LangChain for efficient PDF content extraction.
• Implement Zod for robust API input and LLM response validation.
• Filter retrieved chunks using a `MIN_SIMILARITY_SCORE` (e.g., 0.4).

Topics

• Retrieval-Augmented Generation
• Node.js
• TypeScript
• Qdrant
• OpenAI API
• Large Language Models
• Vector Databases

Code references

• aligorkem/traditional-rag-pipeline

Best for: AI Engineer, Machine Learning Engineer, Software Engineer

Related on AIssential

• What No RAG Tutorial Tells You Before You Start Building — Effective RAG implementation hinges on robust retrieval strategies, not just LLM capabilities, to prevent confident but incorrect answers.
• What is OpenRAG? Unlocking the Future of RAG in Generative AI — RAG remains critical for cost, accuracy, and domain-specific knowledge, even with large LLM context windows.
• Retrieval-Augmented Generation — A Deep Dive into Components — RAG improves LLM accuracy and relevance by integrating external knowledge retrieval with text generation.
• Gemini API File Search: The Easy Way to Build RAG — Google's File Search tool automates RAG infrastructure, enabling multimodal search across text and images with Gemini API.
• What is RAG? — RAG combines LLMs with external data sources to generate more accurate and contextually relevant responses.
• Building a Production-Ready RAG Pipeline Workshop — Building production RAG systems requires advanced techniques beyond basic chunking and embedding.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by LLM on Medium.