About

Hello! My name is Nari (rhymes with “Atari”) and I am a second year PhD student in Carnegie Mellon University’s Machine Learning Department advised by Ameet Talwalkar. I graduated from Harvard in May 2021 with a BA and MSc in Computer Science, where I am grateful to have worked with Finale Doshi-Velez and Hima Lakkaraju.

My research studies methods to help humans understand and evaluate machine learning models. I am interested in interpretability and explainability, human-AI interaction, and methods that can help discover or mitigate systemic failures.

Selected Work

  • Evaluating Systemic Error Detection Methods using Synthetic Images.
    ICML Workshop on Spurious Correlations, Invariance, and Stability, 2022
    Gregory Plumb*, Nari Johnson*, Ángel Alexander Cabrera, Marco Tulio Ribeiro, Ameet Talwalkar
    (arXiV)
  • Use-Case-Grounded Simulations for Explanation Evaluation.
    Under Review.
    Valerie Chen, Nari Johnson, Nicholay Topin*, Gregory Plumb*, Ameet Talwalkar.
    (arXiV)
  • OpenXAI: Towards a Transparent Evaluation of Model Explanations.
    Under Review.
    Chirag Agarwal, Eshika Saxena, Satyapriya Krishna, Martin Pawelczyk, Nari Johnson, Isha Puri, Marinka Zitnik, Himabindu Lakkaraju.
    (arXiV)
  • Learning Predictive and Interpretable Timeseries Summaries from ICU Data.
    AMIA Annual Symposium, 2021
    Nari Johnson, Sonali Parbhoo, Andrew Slavin Ross, Finale Doshi-Velez.
    (arXiV)

* denotes equal contribution.

Teaching

Service

I’ve previously served as elected Co-President of Harvard Women in Computer Science. I also led our Advocacy Council, where I facilitated Harvard CS’s first inclusive teaching training.

I care deeply about mental health advocacy, and previously volunteered as a trained counselor for Indigo Peer Counseling.

Contact

(my first name).(my last name) (at) gmail.com