Introduction
Problem Statement
Technologies and shoutouts
Built by
Architecture Diagram

GeoComply – TikTok TechJam 2025 Submission

🚀 What Inspired Us

Global platforms like TikTok operate across dozens of regions, each with its own strict regulations—from GDPR in Europe to data localization in Brazil. Tracking these rules manually is tedious, error-prone, and slows down feature development. We wanted to turn compliance from a blocker into an enabler by creating a tool that makes regulatory obligations clear, visual, and actionable.

📖 What We Learned

How to combine LLM intelligence with structured data to surface compliance risks effectively.
Handling client-only rendering in Next.js for Leaflet maps without breaking SSR.
The importance of RAG (Retrieval-Augmented Generation) when grounding LLMs with trusted, real-world policies.
Building a hackathon-ready stack that balances speed, usability, and technical depth.

🛠️ How We Built Our Project

Frontend: Next.js with ShadCN UI for a responsive, modern dashboard.
Visualization: Leaflet polygons to highlight compliant vs. non-compliant regions on a world map.
Backend & Storage: Supabase with vector DB integration for RAG.
AI Intelligence: Ollama LLM for compliance flagging, enhanced with Tesseract OCR to extract regulatory text from documents.
Feature Management: CRUD functionality for tracking product features and their compliance requirements.

🤔 Challenges We Faced

Geo-specific complexity: Translating vague regulations into structured, region-based compliance rules.
AI accuracy: Ensuring the LLM provides useful compliance insights without false positives.
Map rendering: Avoiding server-side rendering issues when integrating Leaflet into Next.js.
Time constraints: Balancing ambitious scope (LLM + RAG + OCR + visualization) with hackathon deadlines.