Enter the Brave New World of GenAI with Vector Search
summary
A concise primer that maps generative AI to concrete engineering: explains RAG as a two-part system (a retrieval index plus a generative LLM), shows embeddings are numeric vectors (OpenAI embeddings are 1536‑dim float arrays) stored in vector DBs and queried with approximate nearest‑neighbor math, and lists real tools—LangChain4J, LlamaIndex, Semantic Kernel, Hugging Face, PGVector, Chroma, Milvus and Pinecone—while noting a simple demo returns “hello world” and an early GPT‑2 local model hallucinates and swears. RAG (retrieval-augmented generation) embeddings as 1536‑dim float arrays LangChain4J and PGVector examples
AI-written summary · about
chapters (18)
- 0:00Introduction and welcome
- 1:46Talk agenda overview
- 3:34Speaker introduction and background
- 6:05Brief history of AI and deep learning
- 9:17Generative AI era and chat GPT
- 12:55Players in the GenAI space
- 16:05Modality in generative AI
- 17:51New roles in GenAI development
- 19:57Understanding GPT and LLMs
- 22:29Natural language processing basics
- 23:30Working with large language models
- 26:24Java frameworks for GenAI
- 31:21Retrieval augmented generation
- 34:14Vector databases and similarity search
- 43:03Demo and RAG application example
- 45:51Benefits and challenges of GenAI
- 47:15Resources and contact information
- 49:05Q&A session
related talks
33mEnter the Brave New World of GenAI with Vector Search
48mEnter the Brave New World of GenAI with Vector Search
34mEnter the Brave New World of GenAI with Vector Search
49mEnter the Brave New World of GenAI with Vector Search
23mNode.js and GenAI: Deep Dive into the RAG Architecture
41mGenAI, vector/semantic/hybrid search, RRF, NLP, (...) and other buzzwords
33mVector Databases for GenAI applications
21mHow RAG is Redefining AI Capabilities