PolicyPilot
Link to open source: https://github.com/paarthbhatt/Hackfest_Hackathon_Project
Link to Live Project: https://hackfest-hackathon-project.vercel.app/
PolicyGuard is an AI-powered compliance intelligence platform that automatically ingests free-text PDF policy documents, connects to company databases, and continuously monitors data for policy and regulatory violations.
Core goals:
- Ingest & interpret — Upload PDF policy documents (GDPR, Access Control, Data Retention, IRD) and extract actionable compliance rules using Gemini AI
- Scan & detect — Connect to multiple databases (PostgreSQL, MySQL, MongoDB) and automatically flag records that violate extracted rules, with clear justifications for every violation
- Human oversight — Built-in review workflow with manual override, escalation, and DPO notification flows, with every action logged to an immutable audit trail
- Continuous monitoring — Configurable periodic scans (1h to 72h intervals) that detect new and recurring violations over time
- Remediation — Autonomous Fix Engine generates exact SQL/NoSQL remediation queries per violation, with pre-execution preview and rollback support
- Reporting — Exportable audit reports (CSV, JSON, plain text), per-database compliance scores, trend analysis, and an executive summary dashboard
- AI assistant — Embedded PolicyPilot chat for natural-language queries about violations, policies, and fix priorities
Built for the GDG Hackfest 2.0 dataset, with native support for the Internal Recommendation Document rules (IRD-1 through IRD-5), including bias monitoring for AI recommendation systems under the EU AI Act.
This build was uploaded as a hackathon project
