RAG from Scratch: Build a Knowledge-Powered Chatbot

Board Infinity via Coursera

Go to class Write review

Details

Go to class

Provider

Coursera
Pricing

Paid Course
Languages

English
Certificate

Certificate Available
Effort

5 weeks, 2 hours a week
Sessions

Self-Paced
Level

Intermediate
Subtitles

English

Found in

Overview

Google, IBM & Meta Certificates – 40% Off

One Coursera Plus subscription covers most Professional Certificates on Coursera.

Unlock All Certificates

Learn how to design, build, evaluate, and deploy production-quality Retrieval-Augmented Generation (RAG) systems for real-world enterprise use cases. You’ll start by understanding why LLMs hallucinate and how RAG differs from fine-tuning and long-context strategies. Then you’ll implement a complete naive RAG pipeline from scratch in Python: generating embeddings with OpenAI, computing cosine similarity, building a simple vector store, and wiring up retrieval-augmented prompts that produce grounded, cited answers. From there, you’ll progress to robust ingestion and retrieval pipelines for enterprise documentation. You’ll parse PDFs, Markdown, HTML, and CSVs into unified document objects; compare and tune chunking strategies; enrich chunks with metadata; and work with modern vector databases such as ChromaDB, FAISS, and Pinecone, including metadata filtering, namespaces, and multi-tenant access control. You’ll explore lexical (BM25), semantic, and hybrid retrieval with Reciprocal Rank Fusion, add cross-encoder and LLM-based reranking, and benchmark alternative pipelines using RAGAS metrics and synthetic test sets. Finally, you’ll focus on productionization: building a conversational Streamlit chatbot with conversation memory, query reformulation, and Agentic RAG workflows like query routing, self-corrective retrieval, and fallback strategies. You’ll compare advanced architectures such as Multi-Index RAG and GraphRAG, optimize for quality, latency, and cost through tuning and profiling, and deploy a secure, monitored enterprise RAG chatbot complete with streaming responses, authentication, user feedback loops, and automated re-indexing in the cloud. Disclaimer: This is an independent educational resource created by Board Infinity for informational and educational purposes only. This course is not affiliated with, endorsed by, sponsored by, or officially associated with any company, organization, or certification body unless explicitly stated. The content provided is based on industry knowledge and best practices but does not constitute official training material for any specific employer or certification program. All company names, trademarks, service marks, and logos referenced are the property of their respective owners and are used solely for educational identification and comparison purposes.

Syllabus

RAG Foundations & First Pipeline from Scratch
- In this module, you'll explore the core concepts behind retrieval-augmented generation, understanding why language models hallucinate and how retrieval grounds answers in real evidence. You'll build intuition from first principles, create embeddings, and implement similarity search in raw Python. By completing a naive RAG pipeline without frameworks, you'll connect each technical step to building an enterprise knowledge base chatbot that answers questions with citations.
Document Ingestion, Chunking & Embedding Models
- This module lays the data foundation for a production RAG system. You'll learn to parse PDFs, Markdown, HTML, and CSV files into clean text and unified document objects. You'll apply various chunking strategies that impact retrieval quality and enrich chunks with metadata for filtering and attribution. Additionally, you'll compare embedding models and explore when domain-specific fine-tuning is beneficial for enterprise documentation.
Vector Databases, Hybrid Search & Re-Ranking
- In this module, you'll focus on the retrieval layer that ensures your RAG system finds the right evidence at scale. You'll work with local and managed vector databases, implement metadata filtering and namespace strategies for enterprise control, and enhance retrieval by combining semantic and lexical methods. By the end, you'll build a production-style retrieval pipeline using hybrid search and reranking for precise context delivery.
Generation, Conversation & Advanced RAG Architectures
- This module transitions from retrieval to answer generation, conversational behavior, and advanced RAG orchestration. You'll design prompts that encourage faithful, cited answers, manage context windows and streaming, and address follow-up question challenges with memory and query reformulation. You'll build a conversational chatbot interface and explore advanced architectures like Agentic RAG, Multi-Index RAG, and GraphRAG for complex enterprise workflows.
Evaluation, Optimization & Production Deployment
- In this final module, you'll transform your prototype into a measurable, optimized, and deployable RAG system. You'll evaluate pipeline quality using RAGAS metrics, generate test datasets, and build dashboards to monitor retrieval and generation failures. You'll optimize quality, latency, and cost, then deploy a production chatbot with authentication, monitoring, feedback loops, and continuous improvement practices reflecting real enterprise operations.