FAIR is an AI-assisted investigation and review system developed for the New Jersey Attorney Generals Office Division on Civil Rights (DCR) to streamline case handling, improve documentation consistency, and accelerate investigative workflows.
FAIR is an AI-assisted investigation and review system with three key components:
- Converts video recordings into structured, searchable transcripts
- Significantly improves accuracy by 36% compared to current Microsoft Teams transcription
- Creates consistent documentation that follows standardized formatting
- Generates concise, context-aware summaries based on transcripts
- Structures summaries according to DCR's preferred format for investigative documentation
- Highlights key elements such as allegations, respondent statements, and supporting evidence
- Easy ‘revision’ button to alter summary based on additional context or personal preferences
- Allows users to ask natural language questions about the interview content
- Retrieves targeted insights without requiring full transcript review
- Enables efficient information retrieval without technical search expertise
| Layer | Tools & Frameworks |
|---|---|
| Frontend | Next.js, Tailwind CSS, Azure Static Web Apps |
| Backend | Flask (Python), Azure Container Apps, Docker, Azure Container Registry (ACR) |
| Database | PostgreSQL, Azure Database for PostgreSQL |
| AI/ML | GPT4o Transcribe (OpenAI), GPT-4o (OpenAI) |
| Security & Networking | Secruity through the cloud |
| Infra/DevOps | Docker, Azure Container Registry, Azure Infrastructure (scalable, highly available) |
| Data Privacy | Uploaded documents are not stored; they are deleted immediately after processing |
Implementation Guide
To find the cloud deployment instructions go the deployment section on page 6 and follow the link to the instructions.
| Module | Description |
|---|---|
interview_summarizer.py |
Contains the Prompt for Generating the Summary |
chat.py |
Contains the Prompt for Initializing the AI Assistant |
app.py |
Backend File |
page.tsx |
Frontend File |
requirements.txt |
Contains All the Requirements |
.github/workflows |
Folder Containing GitHub Workflow Files to Push Updates to Azure |
- Documents are processed in-memory only and nothing is saved to a database after the summary is generated
- No document data is persisted once the request completes.
- API is CORS-enabled but can be locked down to authenticated roles via Azure Static Web Apps config.
- Secrets are stored in environment variables.
- Integration with DCR Tech Stack: Host tool on DCR's Azure cloud account and make accessible through NJ BIAS
- User Feedback Mechanism: Add feedback buttons to summary and home pages
- Expand multi-language support for diverse New Jersey communities
- Develop domain-specific models trained on civil rights legal terminology
If you are contributing please follow these steps:
- Fork the repository.
- Create a new branch (
git checkout -b feature/YourFeature). - Make your changes and commit them (
git commit -m 'Add some feature'). - Push to the branch (
git push origin feature/YourFeature). - Open a pull request.
MIT License – see LICENSE.md for details.
Built by Arinjay Singh & Dhruv Reddy Tekulapalli
In partnership with the New Jersey Attorney Generals Office Divison on Civil Rights