Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
GenAI Weak Supervision Labeling
This talk demonstrates generating high-confidence labels from messy enterprise data using LLMs, Python AST validation, and weak supervision for applications like fraud detection.
I’ll demo how we generate structured, high-confidence labels from messy enterprise data using LLMs and weak supervision—without manual annotators.
We use Python ASTs to validate GenAI-generated labeling functions, then combine them with a generative model to produce probabilistic labels. These power downstream models for fraud detection, dynamic pricing, and more.