Hello! My name is Nari (rhymes with "starry") and I am a fourth-year PhD student in Carnegie Mellon University's Machine Learning Department, where I'm fortunate to be advised by Hoda Heidari. I graduated from Harvard in 2021 with a BA and MS in Computer Science, where I previously worked with Finale Doshi-Velez.

My research examines how often-opaque algorithmic systems can be better evaluated and governed. To this end, my work adopts methods from machine learning and human-computer interaction to study AI models and the people they impact.

Email: narij at andrew dot cmu dot edu
Nari's headshot.
Refereed Publications
* denotes equal contribution.
The Fall of an Algorithm: Characterizing the Dynamics Toward Abandonment
Nari Johnson, Sanika Moharana, Christina N. Harrington, Nazanin Andalibi, Hoda Heidari*, Motahhare Eslami*
ACM Conference on Fairness, Accountability, and Transparency (FAccT), 2024
[arXiv] [database] [podcast episode]
Where Does My Model Underperform? A Human Evaluation of Slice Discovery Algorithms
Nari Johnson, รngel Alexander Cabrera, Gregory Plumb, Ameet Talwalkar
AAAI Conference on Human Computation and Crowdsourcing (HCOMP), 2023
ICML Workshop on Spurious Correlations, Invariance, and Stability, 2023
[arXiv] [conference talk] [data & code]
๐ŸŒŸ HCOMP 2023 Best Paper Award (top 1)
๐ŸŒŸ Spotlight Presentation (top 5%) at the ICML SCIS Workshop
Towards a More Rigorous Science of Blindspot Discovery in Image Classification Models
Gregory Plumb*, Nari Johnson*, รngel Alexander Cabrera, Ameet Talwalkar
Transactions on ML Research (TMLR), 2023
ICML Workshop on Spurious Correlations, Invariance, and Stability, 2022
[arXiv] [code: PlaneSpot] [code: SpotCheck]
Use-Case-Grounded Simulations for Explanation Evaluation
Valerie Chen, Nari Johnson, Nicholay Topin*, Gregory Plumb*, Ameet Talwalkar
NeurIPS, 2022
[arXiv] [code]
OpenXAI: Towards a Transparent Evaluation of Model Explanations
Chirag Agarwal, Satyapriya Krishna, Eshika Saxena, Martin Pawelczyk, Nari Johnson, Isha Puri, Marinka Zitnik, Himabindu Lakkaraju
NeurIPS Track on Datasets and Benchmarks, 2022
ICLR Pair2Struct Workshop, 2022
[arXiv] [code]
๐ŸŒŸ Oral Presentation (top 5%) at the ICLR Pair2Struct Workshop
Learning Predictive and Interpretable Timeseries Summaries from ICU Data
Nari Johnson, Sonali Parbhoo, Andrew Slavin Ross, Finale Doshi-Velez
AMIA Annual Symposium, 2021
[arXiv]
๐ŸŒŸ AMIA Student Paper Competition Finalist (top 8)
๐ŸŒŸ Knowledge Discovery & Data Mining Student Innovation Award
Workshop Papers & Pre-prints
Assessing AI Impact Assessments: A Classroom Study
Nari Johnson, Hoda Heidari
NeurIPS Regulatable ML Workshop, 2023
[arXiv]
Teaching
I've greatly enjoyed teaching several classes in college and my PhD program. As an undergrad, I designed and led Harvard Computer Science's first inclusive teaching training. My teaching has been recognized with several honors, including Carnegie Mellon's Machine Learning TA Award, Harvard's Alex Patel Teaching Fellowship, and Harvard's Derek Bok Certificate of Distinction in Teaching (4x).
Service
I care deeply about creating supportive communities in computing. At CMU, I organize my department's Wellness Committee and plan community-building events for women and other under-represented gender minorities. I also participated in TechNights and the AI Mentorship Program. In a past life, I served as elected Co-President and led the Advocacy Council of Harvard Women in Computer Science.

Professional Service: I've served as a reviewer for: ICLR 2024 (Outstanding Reviewer), NeurIPS 2023 (Top Reviewer, Highlighted Workshop Reviewer), CHI LBW 2023.