raul raja databricks meetup scala days 2023 seattle 1080p
summary
An open source JVM library called Seth AI provides a monadic AI block DSL for Scala/Kotlin (published to Maven Central as an alpha) and bindings for Java, Rust, Go and Swift in progress. It talks to OpenAI and Hugging Face HTTP APIs today and plans local model support soon, offers typed error ADT (no response, JSON parse, token limits), prompt templates and composable contexts with vector stores and PostgreSQL backends, can ingest PDFs for chat, and is already used in-house to summarize specs and propose code fixes against Alloy/TLA-style models.
AI-written summary · about
chapters (9)
- 0:00Seth AI open source library
- 2:29Functional AI block design
- 5:02Structured AI responses
- 6:09Prompt templates and chains
- 7:14Context and memory for AI
- 9:02Composing AI blocks and storage
- 10:55Extending library with custom functions
- 12:20Real-world use in verification
- 14:29Q&A
related talks
47mScala: The Unpredicted Lingua Franca for Data Science
45mScala Meets GenAI - Build the Cool Stuff with LLM4S
46mScala on the Move!
02mSCALAR 2023
01h 12mBig Data in Java World: Apache Spark and Scala. Arch and live coding
21mLawful AI
27mNavigating Scala’s AI Landscape
47mScala: The Unpredicted Lingua Franca for Data Science