Using data to serve millions: Government data scientists share insights at UC Berkeley panel

At the National Workshop on Data Science Education hosted at UC Berkeley, educators and practitioners gathered to consider how data science education can prepare students for real-world impact. A panel of government experts offered a window into how those skills can strengthen public systems across California, from protecting families from fraud to strengthening water security and preparing for the state’s clean energy transition. 

Leading this effort is Monica Bobra, appointed as California's principal data scientist in 2023. Working within the Office of Data and Innovation, Bobra is the technical lead for the Data Science Accelerator, a specialized team that partners with agencies across government to address complex operational challenges. 

"Why do we care about data-driven decision making?" Bobra said. "It's because we can get people safe and clean drinking water more reliably. It's because we can get people social services more reliably."

Answers to many of government's most pressing questions are buried within massive quantities of data. Data scientists who can build models and translate findings for nontechnical colleagues help agencies turn that information into action, improving services and outcomes for millions of Californians.

Joaquin Carbonell, research bureau chief at the California Department of Social Services, worked alongside Bobra’s team to combat organized theft of Electronic Benefits Transfer (EBT) funds. 

By building modern data infrastructure, the department analyzed millions of daily transactions. As criminal card-skimming operations came into focus, the department reduced losses from a peak of roughly $21 million per month by 80%, protecting families relying on benefits. 

Similarly, Dan Wang of the State Water Resources Control Board collaborated with the accelerator to build a machine learning model that forecasts how drought conditions could affect drinking water across California. By integrating environmental, geospatial and socioeconomic data, the model allows agencies to identify vulnerable systems before communities lose access to safe drinking water.

The complexity of these problems and limits to resources mean solutions are rarely straightforward. 

“We’ve got to find the sweet spot of technically accurate versus practically possible,” Wang said.

Still, those same challenges create opportunities for data scientists to make immediate and meaningful impacts. 

Alan Jian, an electric generation system specialist at the California Energy Commission and UC Berkeley graduate (B.A. Data Science '22, Master of Information and Data Science '23), uses his data science skills to help California prepare for a clean energy future. 

As the state rapidly electrifies transportation, Jian develops models that forecast how and where electricity demand from electric trucks will grow. Using petabytes of hourly smart meter data, his work helps utilities and policymakers anticipate where grid upgrades will be needed, allowing Calfornia to proactively prepare for increasing demand rather than responding to shortages later.

"We don't want to run into situations where we have rolling blackouts," Jian said. "We want to stay ahead of the ball."

Lucy Andrews’ team at the California Department of Water Resources tackles problems ranging from emergency flood response to habitat restoration. But recently, she led a project aimed at making publicly funded science more accessible and accountable. The department invests tens of millions of dollars annually in scientific research, prompting questions from legislators and the public about what that investment produces.

To answer those questions, Andrews’ team used large language models to inventory and classify the agency’s extensive scientific publications, then built an open-source dashboard and chatbot that allow decision-makers to easily explore research. The tools also ensure the public can access and understand the science funded with taxpayer dollars. 

"Our code bases, in my opinion, belong to the people of California," Andrews said.

For students inspired to use their data skills toward this kind of mission-driven work, panelists offered guidance on navigating the civil service system.

Andrews noted that the title of “data scientist” rarely appears in state job listings. Instead, panelists encouraged exploring other classifications, including information technology specialist or research analyst.  

Because government work often favors adaptable problem-solvers, panelists recommended that students seek opportunities to work with messy, real-world data sets. Wang highlighted Berkeley’s Data Discovery Program, which matches students with organizations to address real-world challenges through data science projects.

Panelists closed by reflecting a shared hope that more students will see government as a career pathway that offers intellectually engaging technical problems whose outcomes are directly felt in people’s daily lives, from access to basic needs to the resilience of essential public systems. 

"What binds all of us here on this panel,” said Bobra, “is that every single data scientist deeply cares about the work that they're doing and the mission of their department."

The annual National Workshop on Data Science Education is a multi-day event series that offers opportunities to learn about data science education and meet fellow educators involved in shaping the discipline at the undergraduate level.