Use your CS/AI/ML coding and hacking skills to help direct cash transfers to poor households in need!
Joshua Blumenstock, Professor
Information, School of
Applications for Spring 2026 are closed for this project.
At the Global Opportunity Lab (gol.berkeley.edu), we view the world’s most pressing policy challenges as opportunities to develop transformative policy solutions. Harnessing state-of-the-art data science, computation, and analytic tools, we generate insights and innovations that promote prosperity and expand opportunity across the globe.
Our lab is currently working on several projects to distribute cash transfers to extremely poor households in several developing countries, including Malawi, Haiti, Afghanistan, and Bangladesh. This work has been covered in the news:
- https://www.bbc.co.uk/programmes/p099r62z -
- https://www.npr.org/sections/goatsandsoda/2021/02/15/966848542/the-pandemic-pushed-this-farmer-into-deep-poverty-then-something-amazing-happene
- https://www.wired.com/story/clever-strategy-distribute-covid-aid-satellite-data/
- https://www.fastcompany.com/90625436/these-new-poverty-maps-could-reshape-how-we-deliver-humanitarian-aid
We have also written several papers about this work:
- https://www.nature.com/articles/s41586-022-04484-9
- https://www.pnas.org/doi/10.1073/pnas.2113658119
- https://www.nature.com/articles/d41586-020-01393-7
- https://www.science.org/doi/10.1126/science.aac4420
Role: Your mission, should you choose to accept it, is to help us improve the algorithms that are used to identify the poorest households in a community, so that they can be "first in line" to receive cash transfers from the partners we support (World Food Programme, GiveDirectly, World Bank, and several government partners in low-income countries).
Specifically, we have developed an open-source code repository, CIDER (https://global-policy-lab.github.io/cider-documentation/intro.html), that we use to convert raw "big" data -- from mobile phone networks, social media, and satellite imagery -- into features that can be used in machine learning algorithms. We are entering a phase of the work where we need to improve this feature engineering process, to ensure that our algorithms are extracting the most useful and interesting information from the raw data.
This semester, URAP apprentices will work together (and perhaps compete against each other?) to try and develop the best features for predicting wealth from mobile phone data. For instance, you could design a set of features that better capture the social network structure of someone's text message network. Or you could come up with a creative approach to capturing the mobility patterns evident in the sequence of cell towers that a person connects to. Or you could come up with a novel approach to graph representation learning. You get the idea.
The point is, you will be writing code (in Python) to process big data (from mobile phone networks), in order to extract informative features from the data. If you are successful, the features you design will be included in a machine learning pipeline that will help direct millions of dollars in cash to ultra-poor households around the world.
Qualifications: All URAP apprentices, irrespective of the specific role or assignment, are expected to be extremely self-motivated, attentive to detail, and meticulous in their approach to data analysis. Apprentices must be able to work independently and be excited to take responsibility and initiative for ensuring their work meets exacting standards of quality.
Specific qualifications for this position include:
- Proficiency in Python (required)
- Experience with large datasets (required)
- Ability to extract relevant information from data and present it clearly, especially through visualization (required)
- Knowledge of statistical/econometric methods (preferred)
- Track record of contributions to projects designed to benefit society
Day-to-day supervisor for this project: Leo Selker, Staff Researcher
Hours: to be negotiated
Related website: https://gol.berkeley.edu
Related website: https://www.jblumenstock.com