user_image

Avni Kothari

I am currently working as a Data Scientist with Jean Feng at UCSF to help research and deploy ML algorithms for Zuckerberg San Francisco General Hospital. My research focuses on safety in machine learning from the perspective of 1) making machine learning models transparent and aligned with human intuition in high stakes settings, and 2) ensuring equitable and fair ML models by detecting individuals subject to preclusion. I recently graduated from my Master's in Computer Science at UC San Diego where I was named a DeepMind Fellow. At UCSD I worked with Berk Ustun and Lily Weng in algorithmic recourse. Previously, I was a software engineer creating web-applications in education and healthcare. Outside of work you can usually find me biking or enjoying the outdoors.

My CV can be found here.

Don't hesitate to reach out if you want to learn more about me or my work!

Papers

  • Concept Bottleneck Models with LLM Priors

    Jean Feng, Avni Kothari, Lucas Zier, Chandan Singh, Yan Shuo Tan

    preprint, 2024

    This work eliminates the need for human-annotated concepts by proposing a novel method to learn concepts by wrapping LLMs within a Bayesian framework. This approach is highly generalizable across various data modalities and allows for rigorous uncertainty quantification despite LLMs being prone to error and hallucinations.

    Prediction without Preclusion: Recourse Verification with Reachable Sets

    Avni Kothari*, Bogdan Kulynych*, Lily Weng, and Berk Ustun

    ICLR (Spotlight), 2024

    Individuals can be assigned predictions that they cannot change through actions on their features. This paper investigates and formalizes scenarios of predictions without recourse. We argue the importance of these scenarios for both model development and recourse detection methods.

    Bayesian Priors From Large Language Models Make Clinical Prediction Models More Interpretable

    Avni Kothari, Daniel J. Bennett, Seth Goldman, Elizabeth Connelly, James D. Marks, Lucas S. Zier, Jean Feng

    AMIA Podium Abstract, 2024

Industry Experience

  • September 2023 – Present

    UCSF, Data Scientist - San Francisco, CA

    Researching and deploying healthcare ML models at San Francisco General Hospital under Jean Feng

    Researching and implementing methods with LLMs to align tabular machine learning models with clinical intuition for model interpretability and reliability

    Researching and evaluating using LLMs in conjunction with Bayesian methods to extract concepts from clinical notes

    Creating, evaluating, and deploying a 30-day all cause readmissions model for use at the hospital

  • Jan 2020 – May 2021

    Edovo, Software Engineer - Chicago, IL

    Designed and developed an educational content platform to handle 700K+ requests per day

    Created a pipeline and nightly job to merge 4 billion rows of user event data in PostgreSQL

    Spearheaded team sessions to improve software development practices and adopt new frameworks

  • Aug 2017 – March 2019

    8th Light, Lead Software Engineer - Chicago, IL

    Developed a diabetes management iOS app to connect patients with diabetic nurse specialists

    Enhanced a Java-based continuous deployment pipeline, seamlessly integrating with internal tools

    Mentored peers and residents through pair programming sessions and code reviews

    Created an HTTP Server in Java without libraries for app deployment

Teaching Experience

  • Fall 2022

    TA for DSC 291 - Interpretability and Explainability in Machine Learning

  • May 2011 – August 2013

    Differential Calculus Tutor

Service

  • Vision 1948 (May 2023 – Present)
  • PenPal for the Incarcerated (September 2020 – Present)
  • Warren Community Garden (August 2021 – Present)
  • UCSF AI4All (July 2024 – July 2024)
  • The Recyclery (August 2018 – May 2021)