Education

2019 - 2024
PhD, Electrical Engineering and Computer Science
University of California, Berkeley
  • Dissertation: Feedback Driven Dynamics in Socio-Algorithmic Systems

2013 - 2017
BSE, Operations Research and Financial Engineering
Princeton University, NJ
  • Summa Cum Laude
  • Certificate in Applications of Computing.
  • Thesis: ”Monotonically Constrained Polynomial Regression: An Application of Sum of Squares Techniques and Semidefinite Programming”.
    • Winner of Procter & Gamble Prize for Outstanding Senior Thesis.
    • Advisers: Amir Ali Ahmadi & Georgina Hall.

Teaching

Fall 2021
CS C281A. Statistical Learning Theory
University of California, Berkeley
  • Graduate Student Instructor
Fall 2020
DATA102/STAT102. Data, Inference, and Decisions
University of California, Berkeley
  • Graduate Student Instructor
  • Recipient of Outstanding GSI Award

Work experience

2024 - present
Researcher Scientist
Meta, Menlo Part, CA

Summer 2022
Student Researcher
Google Research, Mountain View, CA
  • Investigated privacy-accuracy tradeoffs for recommendation systems.
  • Improved performance of current techniques by leveraging non-private side information.

2017 - 2019
Software Development Engineering
Commercial Software Engineering, Microsoft, Redmond, WA
  • Engaged with Azure customers, commercial partners and NGOs to prototype innovative solutions using early release Azure products, ML techniques from recent academic work and open source projects.
  • Project Fizzyo
    • Largest Cystic Fibrosis physiotherapy clinical trial in the UK to date aiming to quantify the impact of physical activity and evaluate the role of gamification in airway clearence treatments.
    • In partnership with Great Ormond Street Hospital developed a data analysis pipeline for the breath pressure sensors attached to airways clearance devices as well as Fitbit activity trackers.
    • Quantified breath and physical activity patterns of the trial participants by processing and featurizing large amount of sensor waveform data.
    • Using unsupervised learning techniques identified physical and respiratory activity clusters correlating with clinical outcomes.
  • Famine Early Action Mechanism
    • Crisis alerting system that provides granular famine risk 6-12 month predictions, used by humanitarian aid institutions to preemptively allocate funds to communities at high risk.
    • Collaborated with World Bank Famine Prevention program to build an interpretability framework that quantifies the magnitude of the risk factors
    • Project under executive sponsorship of Microsoft Philanthropies.
  • Prototyped Reinforcement Learning solutions for industrial scenarios, in partnership with a commercial airline and a Formula 1 race team.

2016
Software Development Intern
Microsoft, Redmond, WA
  • Analyzed the CPU and memory performance of BitFunnel, Bing’s recently open-sourced, document index.
  • Identified inefficiencies in Bing’s matching algorithm and proposed an optimization which was successfully implemented. Currently the end-to-end computation cost per search query have been reduced by 15%.
  • Co-authored the research paper discussing BitFunnel’s performance which was be presented at the SIGIR 2017 and won Best Paper Award.

2015
Software Engineering Intern
Empirasign Strategies LLC, New York, NY
  • Developed an API for programmatic ad buying, thus providing a cheaper alternative to Google AdWords.
  • Converted securities data from unstructured emails and spreadsheets in tabular form.
  • Worked on the automated email feature and enhanced the customization capabilities of email alerts.

Academic Distinctions


Service