AI / ML · LAFAYETTE COLLEGE
Insight Lab
I led a cross-disciplinary team building AI/ML models and GenAI workflows on Azure, using U.S. Census data to generate policy recommendations and automate the analysis behind them.
01What made it hard
The modelling was not the hard part. The hard part was that the data scientists and the policy researchers did not share a vocabulary, and the research questions arrived open-ended, the kind you can discuss for a whole semester without shipping anything.
My job was to turn those into something a team could build against: a defined scope, a timeline, and a repeatable analysis pipeline instead of a series of one-off notebooks.
02What we shipped
A working pipeline on Azure that takes Census data through modelling to drafted policy recommendations, with LLM workflows wired into the analysis step so findings get drafted and reviewed rather than written from scratch.
Every recommendation traces back to its source data, which is the only version of this that is actually useful to a policy audience.
03Built with
- Azure ML
- GenAI
- Python
- Product management
// KEEP LOOKING
