Summary

Scrapes Travis County court records (incl. PDFs) → consolidates to markdown → LLM + viz to show recidivism impact.

Problem Statement

Criminal justice data is fragmented across multiple county portals in inconsistent formats. Without consolidation, it’s impossible to identify patterns or measure the impact of bail reform on public safety.

Details

Crawls multiple county portals to pull full case histories (parties, charges, dispositions, bond terms, timelines) and saves linked PDFs. Normalizes into Markdown/JSON for LLM analysis, then builds per-person timelines and flags “released on bail → new offense” sequences.

Produces individual timelines and aggregate charts highlighting how a small cohort contributes disproportionately to incidents. Designed to extend to additional counties as data sources allow.

Images

What We Learned

Consolidating fragmented court data reveals patterns invisible in individual case records. Per-person timelines make recidivism patterns immediately visible. Automated PDF-to-markdown normalization enables LLM analysis at scale. Visualizing the disproportionate impact of repeat offenders provides data-driven context for policy discussions.