Sandlines

Political campaigns traditionally wait weeks for polling results. At Sandlines I built the in-app polling data pipeline that cut that to hours, up to 14x faster than traditional methods, aggregating, cleaning, and transforming responses with modular Python functions so every stage could be reused as the product evolved. On top of it I designed categorical and time-series visualizations that let campaigns watch support shift while there was still time to act on it.

This layer is where I learned that data engineering is mostly about trust: a fast pipeline nobody believes is worthless.