Jonathan Z. Amar

About me

I am a Senior AI Science Manager at Verily Life Sciences (Alphabet) since March 2021, where I lead a team of AI scientists building generative-AI solutions for healthcare. My work centers on AI agents and the improvement loops (harnesses) that make them reliable for clinical and coaching applications on Verily's Violet platform, and on training foundation models over electronic health records and genomics for downstream predictive tasks. I joined initially on the team's COVID-19 analytics efforts, and previously worked on data science for Onduo, a program for proactive disease management.
Prior to Verily, I earned my Ph.D from the MIT Operations Research Center in 2020, advised by Prof. N. Trichakis. My research interests lie in Online Optimization and approximation algorithms, in Machine Learning coupled with Operations Management applications. I further work on bandit problems, online optimization and exploration-exploitation tradeoffs.
Previously I earned my BS and MS from the Ecole Polytechnique in Applied Mathematics and Computer Science. I've had the pleasure to research at the Technion in Robust optimization.
Passionated about applying data-driven methods to solve complex decision-making problems, I look for challenges to get involved in high-impact projects. For instance, I was lucky to intern within the dispatch team at Uber Inc., optimizing real-time and large-scale systems.

---

Selected Publications in AI & Healthcare

Full list on Google Scholar.

• Integrating Genomics into Multimodal EHR Foundation Models
  with the Verily AI team - submitted to ICML
• STEER: Inference-Time Risk Control via Constrained Quality-Diversity Search
  with E. Yang, J.H. Lee, E. Ye and Y. Jia - submitted to ICML
• AI Agents for Conversational Patient Triage: Simulation-Based Evaluation with Real-World EHR Data
  with S. Rashidian, N. Li et al.
• The complexities of patient-centred conversational artificial intelligence
  with J. Matos, N. Li et al.
• Sleepless Nights, Sugary Days: Creating Synthetic Users with Health Conditions for Realistic Coaching Agents
  Findings of ACL 2025
• A Behavioral Science-Informed Agentic Workflow for Personalized Nutrition Coaching
  JMIR Formative Research 2025
• The Geometry of Queries: Query-Based Innovations in Retrieval-Augmented Generation for Healthcare QA
  Machine Learning for Healthcare (MLHC) 2025
• Improving representativeness in trial recruitment: A data-driven approach
  Contemporary Clinical Trials Communications 2026

---

Earlier Working Papers (Operations Research)

• Demand Estimation and Personalized Assortments from Cross-Store Data for Beer Retailer
  with Vivek Farias, Patricio Foncea and Nikolaos Trichakis
• Distribution Free Algorithms for Network Revenue Mangement
  with Nikolaos Trichakis
  presented at RMP 2019 and INFORMS 2019
• The Second-Price Knapsack Problem: Near-Optimal Real Time Bidding in Internet Advertisement
  with Nicholas Renegar - Best Student Paper Award, MIT Operations Research Center (2020)
  presented at RMP 2019 and INFORMS 2019
• Dynamic Approach to Elliciting Customer Preferences
  with Nikolaos Trichakis
  presented at INFORMS 2018 and 2019

---

Projects

• Infection Modeling for COVID-19 Policy Alliance
  within Covid Alliance
• Deep Reinforcement Learning for 2048
  Class project - Dynamic Programming - MIT
• Constructing Uncertainty Sets from Data: an Unsupervised Learning Perspective
  Class project - Robust Optimization - MIT
• Interval Data Classification with Mu-D Partial information and Geometric Interpretation of Robustness
  with Aharon Ben-Tal and Tamir Hazan
• Most Relevant Path in a Network of Friends
  Research project for ShorTouch
• Estimating Joint Spectral Radius and Applications
  Research project for CMAP - Polytechnique
• Indoor Localization, Optimization of Supply Chain
  Research project for Polestar - Finalist for PSC award at Polytechnique

---

Teaching

• 15.761 Introduction to Operations Management
  MIT, Teaching Assistant (Fall 2017 & Spring 2019)
• 15.734 Introduction to Operations Management for Sloan Executive MBA cohort.
  MIT, Teaching Assistant (Summer 2019)