Creating Context: Loading your life into a vector database
summary
Builds a retrieval‑augmented pipeline that takes a natural question, asks an LLM to generate a GraphQL query, executes it against a vectorized dataset, and then asks the LLM to produce a final answer. Uses Airbyte to pull Strava activities, loads them into weaviate (cloud) which calls OpenAI to create embeddings, and runs the demo in Bun/TypeScript. Setup to ingest and query takes about 20 minutes, works but is fragile: results are non‑deterministic and produce non-deterministic math errors (off by factors up to seven), so rely on API aggregations and slicing for correctness.
AI-written summary · about
chapters (22)
- 0:00Intro
- 0:45Who am I
- 2:04Titles
- 2:38Through Line
- 3:04Agenda
- 3:48Disclaimer
- 4:24Context
- 4:55Retrieval AR
- 6:14Unstructured to Structured
- 6:56Vectors
- 8:13Vector databases
- 9:30Use cases
- 10:16Loading data
- 11:55Setup
- 13:35Thanks Melissa
- 14:12Query for context
- 14:49Why graphql
- 16:42Token limits
- 19:02Demo
- 24:39Conclusion
- 25:36Future thoughts
- 27:24Questions
related talks
45mA gentle introduction to Vector Databases
50mA fun and absurd introduction to Vector Databases
01h 35mTutorial: Build AI-Supercharged RAG Apps with a Vector Database
48mA fun and absurd introduction to Vector Databases
01h 23mVector Databases Demystified: Discover how they work
33mVector Databases for GenAI applications
47mExtendable by Design: Building Generative AI Apps with Postgres and Vector Storage and Azure AI
01h 17mGraph Database and Retrieval Augmented Generation