UA Parking Intelligence
Turning 1,766 student voices into a case the administration could act on.
A full-stack analytics platform I built as SGA Treasurer to give University of Alabama administration structured data on student parking for the first time — replacing anecdotes with evidence.
The problem
Frustration with campus parking was constant, but invisible to administration — there was no structured data, just complaints. SGA needed evidence, not anecdotes, to make a case worth acting on.
Approach
I surveyed 10% of UA's undergraduate permit holders and built a Python ETL pipeline over the Qualtrics API: 1,766 anonymized responses, PII scrubbing, and precomputed metrics exported as JSON artifacts that the dashboard reads directly.
Semantic search over free text
I embedded 2,191 free-text responses with gemini-embedding-001 (768-dim), ran cosine-ranked top-K retrieval per query, and engineered generation prompts that feed retrieved quotes to Gemini with quote-level citations — replacing an earlier keyword-overlap baseline so themes surfaced with receipts.
Results
Findings were presented to UA Parking administration in April 2026 — the first comprehensive, data-led case SGA brought to the table. The Next.js dashboard ships executive stats, an interactive theme explorer, and a chat interface; the public demo runs on synthetic data with identical functionality.