LA3-120

Sessions in LA3-120

Personas of the Labs: A Tool for Studying LLM Behavioral and Belief Signatures

AI/ML & Data Science Intermediate

A practical benchmark for detecting hidden, provider-level alignment signatures in LLMs, the kind that never shows up in standard benchmarks, using relative measurement instead of labeled ground truth.

https://youtu.be/Vo7-JUlIBL0

Extraction Trees: Reliable Structured Data from Messy Text

AI/ML & Data Science Advanced

Single-prompt LLM extraction works about 70% of the time and breaks on documents with more than one event. A layered extraction-tree pipeline, demonstrated on a biomedical benchmark with real ground truth.

https://youtu.be/_mJHRVZ54bg

Reading the Tea Leaves: Query Optimization with Postgres and AI

Data Engineering Intermediate

A practical tour of cost-based query optimizers, how to read EXPLAIN plans, and how LLMs can help diagnose and fix slow queries, demonstrated on Postgres.

https://youtu.be/4z4C2sQ-sqw

When the Clock Is Public: Designing Analytics for High-Stakes Decisions Under Scrutiny

Business Analytics & Reporting Intermediate

Public-consequence analytics (campaigns, government deadlines, crisis response) is a different discipline. The habits that work in low-stakes settings can actively hurt you when the deadline is public.

https://youtu.be/WY9XVA5p44c