AIMS Hackathon - Team Firefly:

• Vishalin Naidoo
• Mia Pham
• Vanya Singla
• Nicolette Mashaba

AIMS Compliance Interrogator

This repository contains Team Firefly's submission for the AIMS Hackathon 2025, focusing on transforming Modern Slavery Statements into interactive, queryable intelligence through AI-powered document analysis.

1. Problem Statement

Modern Slavery Statements are critical transparency tools mandated across Australia, UK, and Canada. However, these documents present significant barriers to effective analysis and action:

• Information Overload: Documents often exceed 100 pages with unstructured narrative content
• Time-Intensive Analysis: Manual review requires hours per document, preventing large-scale analysis
• Lack of Standardization: Inconsistent reporting formats make comparative analysis impossible
• Buried Critical Information: Key compliance details are hidden within dense text
• Static Nature: Existing registries function only as document repositories with no interactive capabilities

This creates an "actionability gap" where valuable transparency data remains locked in unusable formats, hindering stakeholders' ability to effectively combat modern slavery.

2. Objective

To bridge the actionability gap by developing a focused AI-powered platform that enables natural language interrogation of Modern Slavery Statements. Our solution transforms static PDF documents into dynamic, queryable resources that empower NGOs, policymakers, researchers, and companies to:

• Extract specific information instantly without manual document review
• Obtain evidence-based answers with direct source citations
• Access complex compliance data through intuitive natural language queries
• Make informed decisions based on transparent, traceable analysis

3. Solution/Data Use Case Description

Core Solution: Interactive Statement Interrogator

The AIMS Compliance Interrogator implements a Retrieval-Augmented Generation (RAG) architecture that processes uploaded Modern Slavery Statements and enables conversational querying.

Technical Architecture:

1. Document Processing Layer

   • PDF text extraction using .NET libraries
   • Text chunking and preprocessing
   • Vector embedding generation via Python microservice

2. AI Intelligence Layer

   • Semantic search using PostgreSQL with pgvector extension
   • Context retrieval based on query similarity
   • Natural language answer generation via OpenAI API

3. User Interface Layer

   • React TypeScript frontend with Material-UI components
   • Drag-and-drop PDF upload functionality
   • Chat-based query interface with source highlighting

Data Integration:

• Leverages the AIMSQA question set from Project AIMS research as structured knowledge base
• Supports Australian, UK, and Canadian Modern Slavery Statement formats
• Integrates with International Reporting Template standards

Use Case Examples:

• Researcher Query: "Does this statement disclose tier-one supplier information?"
• NGO Analysis: "What grievance mechanisms does the company provide for workers?"
• Policy Review: "How does the company assess modern slavery risks in its operations?"

Future Extensions (Pillars 2 & 3)

Gap Analysis Dashboard: Automated compliance assessment against International Reporting Template with downloadable improvement recommendations.

Industry Intelligence Platform: Aggregated trend analysis across companies, sectors, and jurisdictions with interactive visualizations.

4. Pitch

5-Minute Presentation Video: team-firefly-pitch.mp4

Live Application Demo: demo.mp4

Demo Overview:

1. Problem Introduction: The challenge of analyzing 10,000+ unstructured Modern Slavery Statements
2. Solution Demonstration: Live interrogation of uploaded PDF with natural language queries
3. Three Pillars Showcase: Interactive interrogation, automated gap analysis, and industry trends
4. Technical Innovation: Evidence-based answers with source text highlighting and traceability
5. Impact Vision: Transforming global modern slavery intelligence and stakeholder empowerment

5. Datasets

Location: /datasets

Primary Data Sources

AIMSQA Question Set (AIMSQA Questions.xlsx)

• 200+ structured questions covering modern slavery compliance areas
• Simplified question variants for natural language processing
• Categories covering operations, supply chains, risk assessment, and remediation

Modern Slavery Statement Registries

• Australian Modern Slavery Statements Registry
• UK Modern Slavery Statements Registry
• Canadian Supply Chains Reports
• Sample statements for testing and demonstration

Reference Standards

• International Reporting Template (Levels 1 & 2)
• AIMS.au annotation specifications
• Compliance benchmarking criteria

Data Quality Considerations

• Format Standardization: Automated text extraction handles varied PDF formats and quality
• Language Processing: Embeddings model trained on English business language
• Completeness: AIMSQA questions provide comprehensive coverage of reporting requirements
• Validation: Source text citation ensures answer traceability and verification

6. Project Code

Location: /project

Repository Structure

firefly-aims-interrogator/
├── frontend/                 # React TypeScript UI
│   ├── src/components/      # UI components
│   ├── src/services/        # API integration
│   └── src/types/           # TypeScript definitions
├── backend/                 # .NET 8 Web API
│   ├── Controllers/         # API endpoints
│   ├── Services/            # Business logic
│   ├── Models/              # Data models
│   └── Infrastructure/      # Database context
├── ai-service/              # Python Flask microservice
│   ├── app.py              # Embedding generation service
│   ├── requirements.txt     # Python dependencies
│   └── models/             # ML model configurations
├── database/               # SQLite database files
│   ├── aims-interrogator.db # Main application database
│   └── seed-data/          # Initial data load
├── team-firefly-pitch.mp4  # 5-minute presentation video
├── demo.mp4  # Live application demo
└── docker/                # Container configurations

Key Components

Frontend (/frontend)

• PDF upload and display components
• Chat interface for natural language queries
• Results visualization with source highlighting
• Responsive Material-UI design system

Backend (/backend)

• RESTful API with .NET 8 Web API
• PDF text extraction and processing
• Database integration with Entity Framework Core
• OpenAI API orchestration for answer generation

AI Service (/ai-service)

• Flask microservice for text embedding generation
• SentenceTransformers model integration
• Vector similarity computation endpoints

Database (/database)

• SQLite database for lightweight, file-based storage
• Document and chunk storage with vector embeddings
• AIMSQA question repository
• Optimized similarity search queries

7. Additional docs

Location: /docs

Presentation Materials

• 5-Minute Pitch Slides: Complete presentation deck with speaker notes
• Technical Architecture Diagrams: System design and data flow visualizations
• User Interface Mockups: UI/UX design specifications and wireframes

Demo & Video Content

• Live Application Demo: Full walkthrough of all three pillars
• Feature Demonstrations: Individual component showcases
• Demo Scripts: Detailed walkthrough guides for reproducible demonstrations

Development Documentation

• Setup and Deployment Guides: Step-by-step installation and configuration
• API Documentation: Comprehensive endpoint specifications and examples
• Database Schema: Entity relationship diagrams and table specifications
• Integration Examples: Sample code for third-party integrations

User Guides

• Stakeholder User Manual: How-to guides for NGOs, researchers, and policymakers
• Admin Documentation: System administration and maintenance procedures
• Troubleshooting Guide: Common issues and resolution steps

8. Declaration of Intellectual Property

This project is developed by Team Firefly for the AIMS Hackathon 2025 as a research and demonstration tool.

Open Source Components:

• Utilizes publicly available AIMS research datasets and methodologies
• Builds upon open source libraries and frameworks
• Code will be made available under appropriate open source licensing

Responsible Use Statement: This platform is designed for research, education, and methodological advancement. All analysis is based on publicly available Modern Slavery Statements. The tool does not make judgmental claims about specific companies' compliance or ethical conduct. All outputs are exploratory and require independent expert validation for decision-making purposes.

Data Privacy: User-uploaded documents are processed temporarily for analysis purposes only and are not permanently stored or shared with third parties without explicit consent.

---

Team Firefly - Illuminating Intelligence in Modern Slavery Reporting

AIMS Hackathon 2025 | September 10-17, 2025 This project builds on the open research of Project AIMS (AI against Modern Slavery) by Mila and QUT.
GitHub repository: ai4h_aims-au.

Disclaimers

Computational Resources & Comparative Results

• Describe here the resources used in developing your solution (e.g. GPUs, etc).

No Claims About Companies

This repository and its accompanying models, datasets, metrics, dashboards, and comparative analyses are provided strictly for research and demonstration purposes.

Any comparisons, rankings, or assessments of companies or organizations are exploratory in nature. They may be affected by incomplete data, modeling limitations, or methodological choices. These results must not be used to make factual, legal, or reputational claims about any entity without independent expert review and validation.

Do not use this repository’s contents to make public statements or claims about specific companies, organizations, or individuals.

Terms and Conditions

By submitting this solution to the AIMS Hackathon, our team acknowledges and agrees to abide by the Event’s Terms and Conditions.