Illuminating interdisciplinary pathways to solve real problems.
In a world with critical ethical challenges regarding facts, data, and truth.
Featured Programs
GenAI Workshops

A free five–session, hands-on series for Harvard graduate students and postdocs hosted on Tuesdays from October 7 to November 4, 2025.
AWS Impact Computing Project
A five-part video series about how we’re collaborating with AWS to build the field of impact computing.

HDSI Postdoctoral Fellowships
The HDSI Postdoctoral Fellows work independently over a two-to-three year fellowship with the guidance and partnership of Harvard University faculty.

Harvard Data Science Review
As an open access platform of the HDSI, Harvard Data Science Review (HDSR) aspires to be a global forum on everything data science and data science for everyone.

Research Programs
The HDSI supports research in data science methodology and applications through multiple programs that align with research interests across Harvard Schools.

Industry Engagement
Learn from industry data scientists about the role data science plays in their organization, the methods and techniques being used, and their own career trajectory.
Upcoming Events
Embeddings & RAG in Science
AI in Society: Ethics, Handling Privacy, Bias
Industry Seminar: Alan Hanley, GE Vernova
Environmental Costs & the Future of AI
Causal Seminar: Alex Luedtke
Industry Seminar: Tyler Otto, Reddit
Latest News + Links
- Nutrition Security Under Changing ConditionsAt a rural clinic, a health worker records a child’s weight; at a market, a parent chooses what fish to buy. We’re translating real-world signals into timely guidance that supports nutrition security and public health. Linking climate and environmental exposures to malnutrition in a food-insecurity hotspot Project Lead: Christopher Golden What we’re building What it…
- Newly Funded: Creating a Benchmark for Multi-Step Verified ProgrammingLed by Nada Amin, this project aims to build the first large-scale benchmark for verified programming, uniting program synthesis and formal verification. The team will develop a dataset of 1,000 problems—each paired with metadata, specifications, and automatically checkable solutions in the Dafny language—to evaluate how large language models perform when asked not only to write…
- Climate Decisions, Made Clearer and FasterWhen smoke drifts over a city or storms gather on short notice, people need practical answers, not guesswork. We’re turning climate data and models into guidance communities can use. Climateverse: Making Climate Data Actionable Project Leads: Satchit Balsari; Caroline Buckee What we’re building What it enables Physics-Aware Foundation Models for Extreme Atmospheric Events Project Leads: Petros…
Subscribe
Get your weekly dose of data science.