Vitor Hadad, PhD

Applied Scientist
Edge AI, Devices,
Amazon Lab 126
Resume (PDF)
Google scholar
LinkedIn

About me

I'm an Applied Scientist at Amazon Lab126 and part of the Edge AI team.

My work focuses on training and deploying lightweight multimodal perception AI models that run on Echo devices and support next-generation Alexa. At Amazon I've also contributed to a variety of areas, from building ML algorithms for contactless sleep tracking via radar waves to developing AI/ML tools for synthetic data generation and automatic data augmentation.

Prior to this position, I was a postdoc at the Golub Capital Social Impact Lab led by Prof. Susan Athey at the Stanford Graduate School of Business. My research focused on the design and analysis of adaptive experiments such as bandit algorithms. This work led to collaborations with the World Bank to improve contraceptive uptake in developing countries, increase parking fee resolution in New York City, among others. I was also a core developer on the R/C++ package grf for nonparametric causal inference based on random forests.

I hold a PhD in Economics from Boston College. My advisors were Stefan Hoderlein, Arthur Lewbel and Utku Ünver. My dissertation was about using bandit algorithms to improve the performance of kidney exchange pools in the US.

Journal and conference articles

    Can personalized digital counseling improve consumer search for modern contraceptive methods?

    Susan Athey, Katy Bergstrom, Vitor Hadad, Julian C Jamison, Berk Özler, Luca Parisotto, Julius Dohbit Sama [alphabetically]

    Science Advances | 2023

    Adapting to Misspecification in Contextual Bandits with Offline Regression Oracles

    Sanath Kumar Krishnamurthy, Vitor Hadad, Susan Athey

    PNAS | arXiv | GitHub | 2021

    Off-Policy Evaluation via Adaptive Weighting with Data from Contextual Bandits

    Ruohan Zhan, David A. Hirshberg, Vitor Hadad, Susan Athey

    KDD | arXiv | GitHub | 2021

    Extending Tractable Contextual Bandits Beyond Realizability

    Sanath Kumar Krishnamurthy, Vitor Hadad, Susan Athey

    ICML | arXiv | 2021

    Tractable Contextual Bandits Beyond Realizability

    Sanath Kumar Krishnamurthy, Vitor Hadad, Susan Athey

    AISTATS (poster) | arXiv | 2021

Preprints and others

    Contextual Bandits in a Survey Experiment on Charitable Giving: Within-Experiment Outcomes versus Policy Learning

    Susan Athey, Undral Byambadalai, Vitor Hadad, Sanath Kumar Krishnamurthy, Weiwen Leung, Joseph Jay Williams [alphabetically]

    arXiv | GitHub | 2022

    Shared Decision-Making: Can Improved Counseling Increase Willingness to Pay for Modern Contraceptives?

    Susan Athey, Katy Bergstrom, Vitor Hadad, Julian C. Jamison, Berk Özler, Luca Parisotto, Julius Dohbit Sama [alphabetically]

    Preprint | 2021

    Sufficient Representations for Categorical Variables

    Jonathan Johannemann, Vitor Hadad, Susan Athey, Stefan Wager

    arXiv | GitHub | 2018

    Increasing the uptake of long-acting reversible contraceptives among adolescents and young women in Cameroon

    Susan Athey,Vitor Hadad, Julian Jamison, Berk Özler, Luca Parisotto [alphabetically]

    Pre-analysis plan | 2021

Software

    grf — Causal inference based on generalized random forests

    R/C++ package. Based on Wager, Tibshirani, and Athey (2019, Annals of Statistics)

    GitHub | Docs

    sufrep — Encoding categorical variables via sufficient representations

    R package. Based on Johannemann, Hadad, Athey, and Wager (2018)

    GitHub