How NYC Uses Data to Shape Fairer Laws

No Vendors, No Black Boxes

Alaa’s team builds almost everything in-house. From community voting tools to CRM systems and hearing software, their stack is open source—Postgres, Python, R, Django, React, and more. This isn’t just a budgetary decision. It’s a values-based one.

"Open source ensures reproducibility," he told us. "If we’re going to impact people’s lives, the analysis must be transparent, the software must be inspectable, and the results must be accountable."

That mindset stands in stark contrast to many private-sector deployments of machine learning, where results are often opaque, irreproducible, and unexplainable.

Building Resilience Through Strategic Flexibility

What makes Alaa's team exceptional isn't just what they build—it's how and why they build it. Theirs is a model of civic data work shaped by discipline, restraint, and a clear sense of responsibility.

One example? Despite assumptions that public agencies have easy access to vast datasets, the reality is often the opposite. "Having the right data to perform a great analysis requires the stars to align," Alaa told us. Data is often siloed, inconsistently reported, or unavailable due to legal and political constraints. That makes the work both harder and more important.

In response, Alaa’s team does more than build tools—they educate, enable, and elevate. They conduct internal trainings to help policy staff become smarter consumers of data. They advocate for more disaggregated data, knowing that better questions lead to better decisions.

And then there’s the engineering craftsmanship: one standout is a tool that allows analysts to recalculate population data across nonstandard geographic boundaries. By combining census estimates with building-level data, they can generate accurate demographic breakdowns at the council district level—unlocking insights where raw data alone falls short.

Even their use of AI shows intentionality. Years ago, before the rise of GenAI, the team built a vector-based search tool to check for duplicate legislative proposals. It isn’t flashy, but it is useful. And crucially, it is never allowed to make determinations. It is used as a tool, and the onus of  fact checking the output remains on the end user.

Lessons for the Rest of Us

Alaa didn’t start in public policy. He came from a physics PhD and spent time building machine learning systems at Los Alamos. But when he began working on civic data, he shifted gears—emphasizing statistics over machine learning, and transparency over automation.

Why? Because in high-impact settings, rigor matters more than novelty. And reproducibility builds more trust than velocity ever could.

Talking with Alaa was humbling.  In a time when we're all too willing to hand the reins to algorithms we barely understand, it was refreshing—honestly, grounding—to see someone doubling down on clarity, accountability, and work that invites inspection, not evasion. No smoke, no mirrors. Just good data, good questions, and the kind of transparency that keeps the whole thing honest. That when data is applied thoughtfully, it can do more than just drive better business outcomes.