Zane Shelby
Zane Shelby is a co-creator of InferenceQL, a Clojure/Script-based probabilistic programming extension to SQL for automated Bayesian inference, and currently serving as engineering lead. Zane is also a core developer of Gen.clj, a Clojure/Script implementation of Gen that is part of MIT's OpenGen intended for teaching, research, and embedded enterprise & UX probabilistic programming. Previously, Zane was engineer #2 for US President Obama's 2012 re-election campaign; engineer #3 at Upworthy, where he built one of the web's earliest and fastest growing "attention minutes" measurement systems; an engineer at startups ranging from just ~10 to ~2500 employees; and an early employee of data-driven health insurer Quartet Health, where Zane also served as a platform engineering manager.
Growing Clojure via InferenceQL Synthetic Data Generators
This talk will share a vision for how we hope to grow Clojure’s commercial and civic impact, by reducing risk from data breaches and compliance with privacy regulation, in this era of growing remote work. Our project, InferenceQL, is a new Clojure/Script platform for automated inference that learns to generate accurate synthetic data from a broad range of databases. We will share how Clojure’s distinctive capabilities and values have helped us build InferenceQL, and provide evidence that InferenceQL’s probabilistic programming approach overcomes limitations of neural network approaches to synthetic data. We will also show how InferenceQL can potentially help audit for (and reduce) bias in ML deployments, and raise awareness of DEI issues such as salary inequity in the tech industry. Finally, we will share a demo combining large language models such as GPT3 with InferenceQL, allowing users to query Clojure probabilistic programs in natural language, answering questions in seconds that would otherwise take hours for domain experts and data scientists working together.