Faculty Profile: Prof. Aditi Raghunathan — CMU MLD

Position: Assistant Professor Institution: Carnegie Mellon University, Machine Learning Department Website: https://www.cs.cmu.edu/~aditirag/ Report date: 2026-06-12

Research Focus

ML robustness and safety, distribution shift, adversarial examples, certified defenses, trustworthy ML systems. Core question: how do we build ML systems that are reliable under deployment?

Academic Profile

PhD: Stanford University
NSF CAREER Award
Sloan Research Fellowship
Previously: postdoc at UC Berkeley (with Moritz Hardt / Ben Recht group)

Key Publications

Paper	Venue	Focus
Certified defenses against adversarial examples	NeurIPS/ICML	Provable robustness
Distribution shift and generalization	Various	Out-of-distribution reliability
Safety and robustness in foundation models	2024–2025	Applying robustness to LLMs

Fit with Weijia Zhang

Dimension	Assessment
ML robustness / trustworthiness	✅ Core expertise
LLM safety (foundational)	✅ Active direction
Distribution shift	✅ Strong
NLP agents	❌ Not primary
GUI / VLM agents	❌ Not her focus
RL post-training	❌ Not primary

Verdict

P3 套磁（若做 trustworthy AI 方向）。 Strong researcher but focus is on foundational ML robustness, not agentic AI. Only relevant if Weijia wants to add certified robustness / safety theory angle to work.