TrainSim: A GenAI Restaurant Server Training and Customer Simulation ChatBot
summary
Builds an MVP that turns a dinner PDF into chunked embeddings, hypothetical Q&A and a Neo4j knowledge graph using LangChain RAG, OpenAI for question generation and Neo4j as the vector/graph store. Pipeline extracts sections, ranks ingredients (garlic, mozzarella, tomato top), writes nodes/relationships, and serves a Jeopardy-style Streamlit game for server training. Processing currently takes ~2 minutes per PDF, does not use full graph-RAG retrieval, and the simulation improves feedback scoring though the trainee has not yet taken the official test.
AI-written summary · about
chapters (7)
- 0:00Project overview
- 1:03Designing the solution
- 2:14Tools and architecture
- 4:06Technical implementation
- 7:23Jeopardy game demonstration
- 8:37Project reflection
- 10:07Q&A and closing
related talks
58mDemystifying GenAI: Build a ChatGPT App with Vector Store
25mLets Build A Node.js ChatGPT-Like Bot With AI APIs
48mUsing AI in Software Design: How ChatGPT Can Help With Creating a Solution Architecture
46mUsing AI in Software Design: How ChatGPT Can Help With Creating a Solution Arch.
38mFrom Chat-with-PDF to Quiz-Master
27mBuilding AI applications: Successes and Lessons
48mCreating a Board Game Chatbot with Postgres, AI, and RAG
49mBuilding AI Enabled Web Apps