Jared Donohue

Researcher and Product Data Scientist

About Me

I'm a product data scientist who works at the intersection of research methods and product thinking, using experimentation and causal inference methods to better understand user behavior and measurably improve products. Previously, I spent eight years at Amazon leading engineering and product teams, and more recently, I've worked directly with startup founders to instrument product analytics, influence product strategy, and ship key user-facing features. I'm currently researching causal inference and experimentation methodology with professors at Columbia Business School.

Product

Alexa Plus

Alexa+

Launched innovative audiovisual screen responses for Alexa+, the next-generation of Alexa powered by generative AI. Read more here.

Amazon Echo Show

Echo Show

Increased device interaction and customer satisfaction with Alexa's smart display devices. Read more here.

Alexa Answers Website

AlexaAnswers.com

Led 10x growth in monthly active users through product experiments and marketing expansion. Read more here.

Mayah Design

Mayah Design

Built AI image analysis & product analytics to boost customer satisfaction for Columbia Business School startup: Mayah Design.

Health Evidence Website

HealthEvidence.org (personal project)

Surfaces ClinicalTrials.gov studies and grades them by evidence quality, making it easy to distinguish randomized intervention trials from weaker observational designs at a glance.

Eatopia

Eatopia (personal project)

A nutrition app prototype that uses behavioral science principles to nudge users toward healthier eating.

Research

Using LLMs for Science Guest Lecture

Guest Lecture: Using LLMs for Science

Guest lecture at CUNY Graduate Center for class of 15 PhD and master's students on practical methods and workflows for using large language models in scientific research: PDF

Cloud Seeding Image

Scientific Data Creation & Validation using LLMs

Published "Structured dataset of reported cloud seeding activities in the United States (2000–2025) using an LLM". Built pdf-to-text pipeline with OpenAI integration to extract metadata from 1,000+ scanned NOAA reports, addressing a gap in structured cloud seeding data: Scientific Data (Nature Portfolio)

Cloud Seeding Difference-in-Differences Analysis

Causal Analysis of Cloud Seeding with Difference-in-Differences

Estimated the causal effect of cloud seeding on precipitation using a within-site difference-in-differences design across U.S. seeding sites (2000–2025). Combined the structured NOAA cloud seeding dataset with PRISM and ERA5 precipitation data to compare seeded vs. unseeded months within target and control areas, with results delivered through an interactive Streamlit app: Streamlit App | GitHub

Behavioral Science Image

Behavioral Economics Experiment

Experiment design test for level-k strategic route choices in navigation apps (e.g. Google Maps): PDF

Education

Columbia University

M.S. Data Science, Columbia University

  • Data Science Institute Scholar
  • Columbia Build Lab Intern
  • Selected Coursework: Design & Analysis of Online Experiments, Causal Inference, Statistical Inference & Modeling, Philosophy of Science (audit), User Interface Design, Behavioral Economics
  • External Guest Lecture (CUNY Graduate Center): Using LLMs for Science (PDF)
George Mason University

B.S. Computer Science, George Mason University

  • Computer Science Teaching Assistant and Peer Mentor
  • Selected Coursework: Computing for Scientists, Physics, Intermediate Microeconomics, Advanced Object-Oriented Programming, Mobile Application Development, Rapid Web Application Prototyping