BYOMKESH IntelAI

Explainable Conversational Crime Intelligence Copilot
Built for KSP Datathon 2026 – Challenge 01
Intelligent Conversational AI for KSP Crime Database

📖 About The Project

BYOMKESH IntelAI is a state-of-the-art Explainable Crime Intelligence Copilot designed to modernize criminal investigations for the Karnataka State Police (KSP). It bridges the gap between massive, unstructured FIR data and actionable intelligence by transforming conventional database querying into an intelligent, secure conversational interface.

Rather than relying on complex SQL queries or rigid dashboards, investigators can interact with the system using natural language (Text or Voice). The platform securely interprets investigative intent, retrieves verified records from the official FIR database, analyzes criminal networks, and produces 100% explainable insights with complete audit transparency.

Our Vision: "Transform Crime Data into Actionable Intelligence without compromising on Trust, Explainability, or Security."

🎯 The Challenge: KSP Datathon 2026 (Challenge 01)

The State Crime Records Bureau (SCRB) maintains critical crime records from over 1,100 police stations across Karnataka. Existing investigative workflows are heavily dependent on manual queries and static dashboards, creating a bottleneck for deep investigation, pattern discovery, and real-time intelligence gathering.

The BYOMKESH Solution: An AI Copilot that acts as a digital investigative partner. It not only fetches records but actively assists officers in identifying hotspots, drawing network connections, and forecasting trends—all backed by zero-hallucination verification.

⚡ Key Capabilities & Features

🧠 1. Explainable AI (XAI) Investigation Copilot

Conversational Queries: Supports advanced natural language filtering (e.g., "Show me recent chain snatching cases in Mysuru involving repeat offenders").
Voice-to-Query: Hands-free operation with multilingual voice recognition (English, Kannada, Hindi).
Zero-Hallucination Engine: Every AI response is strictly grounded in the FIR database.
Transparent Reasoning: Provides an "Explainability Audit" for every answer, showing exactly which records, filters, and fields were queried to formulate the response.

🕸️ 2. Advanced Crime Intelligence

Criminal Network Visualization: Instantly maps relationships between FIRs, repeat offenders, and accomplices using D3.js force-directed graphs.
Geospatial Hotspot Analytics: Heatmap overlays highlighting crime density across Karnataka districts.
Predictive Trend Analytics: Time-series analysis forecasting potential crime spikes based on historical FIR dates.
Dynamic Evidence Explorer: A highly robust, fast data grid for filtering, sorting, and examining detailed FIR case files.

🛡️ 3. Enterprise-Grade Security

Role-Based Access Control (RBAC): Strict permissions segregating Admin, DSP, and Inspector access levels.
LLM Isolation: The LLM never touches the database directly. It parses queries into a strict Abstract Syntax Tree (AST), which the backend validates before secure execution.
Cryptographic Audit Trails: Every query is logged with a SHA-256 hash chain to ensure non-repudiation and legal compliance.

📄 4. Professional Reporting

One-Click Export: Instantly generate professional, timestamped PDF Investigation Reports encapsulating the query, resulting data, and AI summary.

🏗️ System Architecture

flowchart TD
    %% Styling Definitions
    classDef default fill:#1e293b,stroke:#475569,stroke-width:2px,color:#f8fafc;
    classDef user fill:#0284c7,stroke:#bae6fd,stroke-width:3px,color:#fff;
    classDef client fill:#0f172a,stroke:#3b82f6,stroke-width:2px,color:#fff;
    classDef ai fill:#4c1d95,stroke:#c4b5fd,stroke-width:2px,color:#fff;
    classDef db fill:#065f46,stroke:#6ee7b7,stroke-width:2px,color:#fff;
    classDef security fill:#991b1b,stroke:#fca5a5,stroke-width:2px,color:#fff;

    %% Nodes
    Officer([Investigator / Officer]):::user
    
    subgraph Frontend ["Client Tier (React + Vite)"]
        UI["Web Dashboard Interface"]:::client
        Voice["Voice Recognition (EN, KN, HI)"]:::client
        Viz["Visualizers (D3, Leaflet, Recharts)"]:::client
    end
    
    subgraph CoreEngine ["AI Engine"]
        NLP["Natural Language Processor"]:::ai
        Intent["Intent Classifier & Entity Extractor"]:::ai
        ASTBuilder["AST Query Compiler"]:::ai
    end
    
    subgraph SecurityLayer ["Zero-Trust Boundary"]
        RBAC["Role-Based Access Control"]:::security
        Validator["AST Sandbox & Policy Validator"]:::security
        Audit["Cryptographic Audit Logger"]:::security
    end
    
    subgraph BackendStore ["Catalyst Infrastructure"]
        DAL["Data Access Layer"]:::db
        FIRDB[("Official KSP FIR ER Database")]:::db
        ExplainEngine["XAI Explanation Generator"]:::ai
    end

    %% Workflow Connections
    Officer -->|Text/Voice| UI
    Officer -->|Microphone| Voice
    Voice --> UI
    
    UI -->|Raw Query| RBAC
    RBAC -->|Authorized Request| NLP
    
    NLP --> Intent
    Intent --> ASTBuilder
    
    ASTBuilder -->|Generated AST JSON| Validator
    Validator -->|Validated Secure Query| DAL
    Validator -.->|Log Query Hash| Audit
    
    DAL --> FIRDB
    FIRDB -->|Raw Records| ExplainEngine
    
    ExplainEngine -->|Evidence + Summary| UI
    ExplainEngine -.->|Log Result Hash| Audit
    
    UI -->|Render Data| Viz
    Viz -->|Export| PDF["Generate PDF Report"]

🔍 Investigation Workflow

sequenceDiagram
    autonumber
    actor Officer as 👮 Investigator
    participant UI as Copilot Interface
    participant NLU as AI Intent Engine
    participant Sec as Security Sandbox
    participant DB as FIR Database
    participant XAI as Explainability Engine

    Officer->>UI: "Show repeat offenders in Mysuru for Robberies"
    activate UI
    UI->>NLU: Send natural language query
    
    activate NLU
    NLU->>NLU: Detect Intent: NETWORK_SEARCH
    NLU->>NLU: Extract Entities: [Mysuru, Robbery, Repeat_Offender]
    NLU-->>Sec: Generate Abstract Syntax Tree (AST)
    deactivate NLU
    
    activate Sec
    Sec->>Sec: Validate AST against Security Policies
    Sec->>DB: Execute mapped secure query
    deactivate Sec
    
    activate DB
    DB-->>XAI: Return structured FIR records & entities
    deactivate DB
    
    activate XAI
    XAI->>XAI: Synthesize verification logic
    XAI->>XAI: Generate Evidence-backed Summary
    XAI-->>UI: Return Summary + Audit Details + Data
    deactivate XAI
    
    UI->>UI: Render Network Graph & Data Grid
    UI-->>Officer: Display Interactive Results & Explanation
    deactivate UI

💻 Technology Stack

Category	Technologies
Frontend Framework	React 19, TypeScript, Vite
Styling & UI	Tailwind CSS, Lucide Icons, Custom Glassmorphism
Visualization	D3.js (Networks), Leaflet (Maps), Recharts (Trends)
Backend / Serverless	Node.js, Zoho Catalyst Functions
Database	Zoho Catalyst Data Store (Mocked via TypeScript Synthetic Generator)
AI / NLU	Custom AST Parser, Intent Router, Regex Pipelines
Security	SHA-256 Hashing, RBAC, AST Sandboxing

🚀 Getting Started (Local Development)

Prerequisites

Node.js (v18 or higher)
npm (v9 or higher)

Installation

Clone the repository:

git clone https://github.com/your-org/BYOMKESH-IntelAI.git
cd BYOMKESH-IntelAI

Install dependencies:
```
npm install
```
Generate the Synthetic KSP Database: (This creates a localized database of 500+ highly realistic, randomized FIR records across Karnataka)
```
npm run generate-data
```
Start the Development Server:
```
npm run dev
```
Access the Terminal: Open http://localhost:5173 in your browser. Click Access Investigator Terminal to launch the secure dashboard.

📁 Repository Structure

BYOMKESH-IntelAI/
├── public/                 # Static assets
├── server/
│   ├── schema/             # Synthetic FIR data generator (generateSyntheticData.ts)
│   └── functions/          # Zoho Catalyst backend functions 
├── src/
│   ├── components/         # React Components
│   │   ├── auth/           # RBAC Gateway
│   │   ├── chat/           # Conversational Interface & AI Engine
│   │   ├── dashboard/      # Main Application Shell
│   │   └── visualization/  # D3/Leaflet/Recharts Visualizers
│   ├── context/            # React Context (Auth, Language, State)
│   ├── services/           # NLU, PDF Reporting, and Mock APIs
│   ├── types/              # TypeScript Interfaces (AST, Database Schema)
│   └── utils/              # Cryptographic Hashing, Translators
├── package.json
├── tailwind.config.js
└── README.md

🔮 Future Enhancements

Real-time FIR Synchronization: Direct live sync with the SCRB mainframe.
Advanced Link Prediction: Machine learning models to predict future accomplices based on historical MOs.
Cross-State Intelligence Sharing: API federation for analyzing inter-state crime syndicates.
Mobile Field App: A lightweight PWA version for officers on active patrol.

🤝 Team BYOMKESH

Inspired by the legendary Indian detective Byomkesh Bakshi, our team merges classical investigative reasoning with modern Artificial Intelligence to build a trustworthy, transparent digital intelligence platform.

Built for KSP Datathon 2026
Building Trustworthy AI for Smarter Investigations

Name		Name	Last commit message	Last commit date
public		public
server		server
src		src
.gitignore		.gitignore
Police_FIR_ER_Diagram.pdf		Police_FIR_ER_Diagram.pdf
README.md		README.md
index.html		index.html
package-lock.json		package-lock.json
package.json		package.json
postcss.config.js		postcss.config.js
tailwind.config.js		tailwind.config.js
tsconfig.json		tsconfig.json
tsconfig.node.json		tsconfig.node.json
vite.config.ts		vite.config.ts

Navigation Menu

Repository files navigation

BYOMKESH IntelAI

📖 About The Project

🎯 The Challenge: KSP Datathon 2026 (Challenge 01)

⚡ Key Capabilities & Features

🧠 1. Explainable AI (XAI) Investigation Copilot

🕸️ 2. Advanced Crime Intelligence

🛡️ 3. Enterprise-Grade Security

📄 4. Professional Reporting

🏗️ System Architecture

🔍 Investigation Workflow

💻 Technology Stack

🚀 Getting Started (Local Development)

Prerequisites

Installation

📁 Repository Structure

🔮 Future Enhancements

🤝 Team BYOMKESH

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

BYOMKESH IntelAI

📖 About The Project

🎯 The Challenge: KSP Datathon 2026 (Challenge 01)

⚡ Key Capabilities & Features

🧠 1. Explainable AI (XAI) Investigation Copilot

🕸️ 2. Advanced Crime Intelligence

🛡️ 3. Enterprise-Grade Security

📄 4. Professional Reporting

🏗️ System Architecture

🔍 Investigation Workflow

💻 Technology Stack

🚀 Getting Started (Local Development)

Prerequisites

Installation

📁 Repository Structure

🔮 Future Enhancements

🤝 Team BYOMKESH

About

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages