Joseph Ramsey

Timeline

1. 1986–2001

   Mathematics and a philosophy doctorate

   He earns an MS in mathematics and philosophy at Indiana University of Pennsylvania in 1986 and a certificate in Chinese philosophy and language at Nanjing University in 1988. From 1990 to 1997, he works as a teaching assistant in the University of California, San Diego’s Department of Philosophy. He completes a philosophy PhD at UC San Diego in 2001.

2. 1997–2006

   Research computing at Carnegie Mellon

   In 1997, he joins Carnegie Mellon University’s Department of Philosophy as a research programmer. He becomes research programmer and director of computing in 2000, director of computing in 2002, and special faculty and director of research computing in 2006.

3. 2006

   The Conservative PC algorithm

   In 2006, he coauthors “Adjacency-Faithfulness and Conservative Causal Inference” with Jiji Zhang and Peter Spirtes. The paper introduces Conservative PC, which checks orientation-faithfulness during the step that assigns directions to causal links.

4. 2015

   Causal search with a million variables

   In 2015, he publishes “Scaling up Greedy Causal Search for Continuous Variables.” The work describes an optimized Greedy Equivalence Search that solves a sparse model with one million variables and 1,000 samples in 18 hours on a supercomputer node at the Pittsburgh Supercomputing Center.

5. By 2018

   Mapping causal connections in the brain

   By 2018, he builds brain-network simulations on the Pittsburgh Supercomputing Center’s Bridges supercomputer. Working with Clark Glymour and colleagues in the Center for Causal Discovery, he helps test Fast Greedy Equivalence Search on simulated networks with one million voxels; the team also uses fMRI data from 51,000 voxels.

6. 2020

   Tetrad receives a SAIL award

   In 2020, Tetrad, developed by Ramsey with Peter Spirtes, Clark Glymour and Richard Scheines, receives the Super Artificial Intelligence Leader (SAIL) award at the World Artificial Intelligence Conference in Shanghai. The platform is one of four projects selected from more than 800 nominees.

7. 2023

   BOSS and grow-shrink trees

   In 2023, he coauthors NeurIPS research that introduces Best Order Score Search (BOSS) and grow-shrink trees for learning directed acyclic graphs. BOSS searches over orderings of variables, and the team tests it on simulated and clinical resting-state fMRI data.

8. 2025

   Targeted tests for hidden variables

   In 2025, he coauthors “Efficient Latent Variable Causal Discovery: Combining Score Search and Targeted Testing” with Bryan Andrews and Peter Spirtes. The paper introduces FCI Targeted-Testing (FCIT), which uses BOSS to guide conditional-independence tests for causal discovery with hidden variables.

9. As of October 2026

   Computing and CLeaR at CMU

   As of October 2026, he is special faculty and director of research computing in CMU’s Department of Philosophy and helps lead the CLeaR-Aether group. His work includes Tetrad’s causal-search software and AProS with Wilfried Sieg, which combines a proof generator with a tutor that helps students complete logic problems.