Retrieval Augmented Generation (RAG) Bootcamp - From Vector Search to Production-Grade Architecture

• Registration: Open until March 29, 2026
• Course Dates: April 7 & 9, 2026
• PDH: 16
• Price: $1,249
• Required:
  • Access to laptop computer
  • Basic Python literacy (can read and modify code)
  • Willingness to learn by doing
  • Google account (for Colab access)

• Helpful but Not Required:
  • Familiarity with numpy/pandas
  • Understanding of machine learning concepts
  • Experience with APIs or transformers library

• Not Required:
  • Deep learning expertise
  • NLP background
  • Information retrieval knowledge
  • Production deployment experience
  • Paid API keys or services (everything is open-source!)

• This bootcamp is ideal for:
  • Software engineers exploring AI/ML applications
  • Data scientists integrating LLMs into workflows
  • Technical leaders evaluating RAG solutions
  • Researchers in information retrieval and NLP
  • Anyone building document-based AI systems

• This bootcamp is especially valuable if you:
  • Want to understand RAG internals, not just use high-level APIs
  • Are building custom RAG solutions requiring fine-grained control
  • Want hands-on experience with both primitives and frameworks

• Location: 2127 Innerbelt Business Center Drive, St. Louis, MO 63114
  
  • Directions 
  • Also delivered live online via Zoom

This intensive workshop provides a technical foundation in Retrieval-Augmented Generation (RAG) using a "Primitives-First" methodology.

We reject the "magic wrapper" approach. Instead, we use progressive architecture:

Day 1 — The Mechanics (Build from Scratch):
 You will build retrieval engines and RAG pipelines using raw Python libraries (numpy, FAISS, transformers). Every line of code will be transparent. You'll work with the Vaswani dataset (11,429 IR research abstracts with 93 queries) to understand how semantic search actually works—no frameworks, no abstraction layers.

Day 2 — The Production (Orchestrate at Scale):
 You will scale to production using LlamaIndex on the BEIR Programmers benchmark (32K StackExchange programming posts with 876 queries). You'll see how frameworks simplify complexity while building a technical Q&A system that answers real programming questions.

Technology Stack

We use a specific toolset for each stage of learning:

Day 1: The Primitives (White-Box)

Purpose: Understand every component by building from scratch

• Embeddings: sentence-transformers (all-MiniLM-L6-v2)
• Vector Operations: numpy for manual similarity calculations
• Indexing: FAISS for vector search
• LLM: Llama 3.2 (3B) via transformers library
• Evaluation: PyTerrier for IR metrics (MRR, NDCG, MAP)
• Dataset: Vaswani corpus
  • 11,429 scientific abstracts from information retrieval research
  • 93 queries with complete relevance judgments
  • Built into PyTerrier (zero setup)
  • Domain: Learning RAG by searching RAG papers

Day 2: The “Production” Stack

Purpose: Scale to production with orchestration frameworks

• Orchestration: LlamaIndex for pipeline management
• LLM Integration: Same Llama 3.2 through LlamaIndex
• Data Loading: ir_datasets (standard IR benchmark library)
• Dataset: BEIR CQADupStack Programmers
  • 32,000 StackExchange programming posts
  • 876 queries (real programming questions)
  • 1,700+ relevance judgments (qrels)
  • Domain: Technical Q&A from StackExchange
  • Zero preparation needed (built into ir_datasets)