Sonia Phene

I'm currently a researcher at OpenAI, where I've worked on GPT-4o and SearchGPT. I was previously at Google Search, where I launched Search Generative Experience, and Google Brain, where I worked on health AI applications. I believe in using machine learning to positively impact the world.

Experience

OpenAI

April 2024 - Present

  • GPT-4o core contributor (specifically making it available for free users!) - technical report
  • SearchGPT (playing with cool new post-training techniques) - blog post

Google

August 2015 - March 2024

Team Lead and Manager on Search Generative Experience (SGE)

  • Search Generative Experience-focused: I led a team within Google Search owning e2e vertical experiences for some of Search's most vulnerable users.
  • Launched LLM-powered authoritative answers mechanism, helping Indian citizens access government benefits programs
  • Helped hire and build out the Google DC site, as my org was established there

Engineer in Google Brain focused on Health AI Research

  • Google Brain: I worked on Machine Learning applications on the Health's Ophthalmology team, publishing papers on computer vision tools to aid doctors in eye disease diagnoses
  • Trained CNNs to predict eye disease risk and deployed worldwide, screening 250k+ patients
  • Deployed models in India and Thailand for screening in low-resource areas

Publications & Patents

Google Scholar Profile

    A few selected highlights
  • o3 and o4-mini Reasoning Models Release
    Core Contributors (including Sonia Phene)
    OpenAI, 2025
  • GPT-4o Technical Report
    Core Contributors (including Sonia Phene)
    OpenAI Technical Report, 2024
  • Using Machine Learning-Based Trait Predictions For Genetic Association Discovery
    C McLean, B Alipanahi, J Cosentino, Sonia Phene, A Carroll
    US Patent Office, 2022
  • Improving Medical Annotation Quality to Decrease Labeling Burden Using Stratified Noisy Cross-Validation
    Joy Hsu & Sonia Phene, et al.
    NeurIPS Workshop, 2020
  • Deep learning and glaucoma specialists: The relative importance of optic disc features to predict glaucoma referral in fundus photographs
    Sonia Phene, R. C. Dunn, Naama Hammel, Y. Liu, J. Krause, N. Kitade, et al.
    Ophthalmology, 2019
  • Deep learning for the detection of diabetic retinopathy in a nationwide screening program
    Paisan Ruamviboonsuk, Jonathan Krause, et al. (including Sonia Phene)
    Nature Partner Journal: Digital Medicine, 2019

Education

Brown University

Bachelor of Science in Computer Science with Honors, Magna Cum Laude, GPA: 4.0/4.0

Providence, RI (2011 - 2015)

Undergraduate Experience

Zynga ML Intern

Customized level difficulty using KNN techniques (Summer 2014)

Brown University Computer Science Capstone

Parallelization of Variational Inference for Bayesian Nonparametric Topic Models (2014-2015)

Head Teaching Assistant for CS15

Object-oriented programming introductory class of over 300 students. Oversaw staff of 30+ fellow TAs.

Mentorship Contact Info

I love supporting and uplifting others!

If you're a student looking for guidance or advice in tech, machine learning, or career decisions, I'm excited to hear about it and support you!

My passion for enabling the next generation greatly exceeds the amount of time I have for it, so I make best effort attempts to respond. I do receive a lot of inbound, so if you're reaching out, it helps a lot to include a bit about yourself and what kind of advice you're looking for.

You can reach me at mentoring with sonia [at] gmail [dot] com.

Please only contact me at this address if you're a student or new grad. Others should please use my LinkedIn.