Democratizing Access to Justice Through Artificial Intelligence
Mission Statement
India has over 50 million pending court cases. Millions of citizens cannot afford legal counsel. Nyay Setu bridges this gap by putting an AI-powered legal assistant, end-to-end case management, and secure virtual courts in the hands of every Indian — entirely free of charge.
- Why Nyay Setu?
- Key Features
- Tech Stack
- System Architecture & Data Flow
- Quick Start
- Documentation
- Contributing
- Contributors
- License
The Indian judiciary faces a systemic crisis that affects hundreds of millions of citizens:
| Problem | Scale |
|---|---|
| Pending court cases | 50+ Million |
| Average time to resolve a civil case | 10–15 Years |
| Citizens unable to afford legal representation | Hundreds of Millions |
Nyay Setu is built to address this directly. The platform removes the three biggest barriers to legal access — cost, complexity, and distance — by digitizing the entire judiciary workflow and placing an AI legal assistant at every citizen's fingertips.
AI Legal Assistant (Vakil Friend) A conversational AI companion powered by Groq's Llama 3.1 that helps citizens understand their legal rights, navigate case filings, review documents, and get real-time answers in plain, accessible language.
Role-Based Dashboards Secure, tailored portals for every user type in the legal ecosystem: Litigants, Lawyers, Judges, Police, and Administrators — each with a workflow designed around their specific responsibilities.
End-to-End Case Management Full case lifecycle support from initial filing to final order, including a case diary, hearing timelines, document management, and status tracking.
Evidence Vault Structured digital evidence uploads with SHA-256 hash verification, creating a tamper-proof and legally admissible record for every case.
Digital FIR Handling Police can upload FIRs digitally. The AI instantly generates a structured summary and draft charge sheet, significantly reducing manual paperwork and processing time.
Virtual Courtrooms Native WebRTC-based video conferencing integrated directly into the case timeline for secure, seamless remote hearings — no third-party applications required.
Secure Authentication JWT-based stateless authentication with Spring Security, role-based access control, and multi-layer request filtering.
nyay-setu-working/
│
├── frontend/ # React + Vite frontend application
│ └── nyaysetu-frontend/ # Main frontend source code
│ ├── src/
│ │ ├── components/ # Reusable UI components
│ │ ├── pages/ # Application pages/routes
│ │ ├── services/ # API service handlers
│ │ ├── hooks/ # Custom React hooks
│ │ ├── store/ # Zustand/global state management
│ │ ├── utils/ # Helper and utility functions
│ │ └── assets/ # Images, icons, and static assets
│ ├── public/ # Public static files
│ └── package.json # Frontend dependencies and scripts
│
├── backend/ # Spring Boot backend services
│ └── nyaysetu-backend/
│ ├── src/main/java/ # Main backend application source
│ ├── src/main/resources/ # Configuration files and resources
│ ├── src/test/ # Backend test cases
│ └── pom.xml # Maven configuration
│
├── nlp-orchestrator/ # FastAPI-based NLP orchestration service
│ ├── main.py # Main FastAPI application entry point
│ ├── services/ # NLP and AI processing services
│ ├── routes/ # API route handlers
│ ├── models/ # AI/NLP models and schemas
│ └── requirements.txt # Python dependencies
│
├── docs/ # Project documentation
│ ├── architecture/ # Architecture and system design docs
│ ├── setup.md # Setup and installation guide
│ └── additional-guides/ # Supporting documentation
│
├── assets/ # README banners, screenshots, and assets
│
├── .github/ # GitHub workflows and community files
│ ├── workflows/ # CI/CD GitHub Actions workflows
│ └── ISSUE_TEMPLATE/ # GitHub issue templates
│
├── CONTRIBUTING.md # Contribution guidelines
├── README.md # Main project documentation
├── LICENSE # Open-source license
└── SYSTEM_DOCUMENTATION.md # API and system documentation
| Directory | Purpose |
|---|---|
frontend/ |
Contains the React frontend application and UI components |
backend/ |
Handles APIs, authentication, database access, and business logic |
nlp-orchestrator/ |
Manages AI-powered legal assistance and NLP workflows |
docs/ |
Contains setup guides, architecture diagrams, and technical documentation |
assets/ |
Stores README images, banners, and other static assets |
.github/ |
GitHub workflows, templates, and repository automation |
Nyay Setu's architecture is a microservices ecosystem structured into three tiers:
- Frontend Tier (React + Vite): Renders role-based interactive dashboards, provides client-side WebRTC capabilities for virtual hearings, handles audio recording via browser APIs, and features a dynamic 3D Hologram Avatar component.
- Core Backend Tier (Spring Boot): Manages case creation, secure session handling, user credentials, document verification vaults (SHA-256 hashed), Bhashini Indic language services integration, and routes conversational requests.
- NLP & AI Tier (FastAPI): Coordinates advanced legal reasoning pipelines, decompiles multi-part queries, executes parallel web research on Indian Kanoon databases, synthesizes citations, and converts final summaries to Hinglish dialogue.
graph TD
%% Define Nodes
subgraph FrontendApp [Frontend Tier: React / Vite]
UI[User Dashboard & Chat UI]
AV[3D Hologram Avatar]
RTC[WebRTC Client]
end
subgraph BackendApp [Backend Tier: Spring Boot]
Auth[Security & JWT filter]
VF_Svc[Vakil Friend Service]
Doc_Svc[Evidence Vault Service]
Bhash_Svc[Bhashini Translation API Client]
RAG_Proxy[RAG Proxy Client]
end
subgraph NLP_AI [NLP & AI Tier: Python / FastAPI]
NLP_Orch[NLP Orchestrator: Port 8001]
L_GPT[LawGPT Service: Port 8001]
FAISS[(FAISS Vector Database)]
end
subgraph DatabaseTier [Data & Storage Tier]
DB[(PostgreSQL Database)]
Vault[Tamper-Proof Storage]
end
subgraph ExtAPIs [External Integrations]
Groq[Groq API: Llama 3.1 & 3.3]
Gemini[Google Gemini API: 1.5 Flash]
Kanoon[Indian Kanoon API]
Bhashini[Bhashini National Translation Service]
Ollama[Local Ollama: llama3/gemma3]
end
%% Interactions
UI -->|1. REST Request / Auth| Auth
UI -->|2. Direct SSE Stream /research/deep| NLP_Orch
AV -->|Lip-Sync Animation| UI
Auth --> VF_Svc
VF_Svc -->|Save sessions / Case diary| DB
Doc_Svc -->|Write metadata / SHA-256 hash| DB
Doc_Svc -->|Tamper-proof evidence| Vault
VF_Svc -->|Retrieve legal context| RAG_Proxy
RAG_Proxy -->|POST /context| L_GPT
L_GPT -->|FAISS Semantic Index Lookup| FAISS
VF_Svc -->|Translate to Indic / ASR| Bhash_Svc
Bhash_Svc -->|Direct HTTP POST| Bhashini
VF_Svc -->|Llama-3.1 Chat Completion / Fallback| Groq
VF_Svc -->|Local AI Fallback| Ollama
NLP_Orch -->|Decompose & Route| Groq
NLP_Orch -->|Deep Reasoning| Gemini
NLP_Orch -->|Search Judgments| Kanoon
NLP_Orch -->|Hinglish Dialogue Translation| Groq
The system supports two primary request handling paradigms. The sequence diagram below shows both the standard conversational filing path (JWT-secured REST via Spring Boot) and the advanced deep legal research pipeline (direct client SSE connection to Python NLP-Orchestrator).
sequenceDiagram
autonumber
actor User as Citizen / Litigant
participant FE as Frontend (React UI)
participant BE as Backend (Spring Boot)
participant BHA as Bhashini Translation API
participant GPT as LawGPT Service
participant NLP as NLP Orchestrator
participant DB as PostgreSQL
participant AI as AI Services (Groq / Gemini / Ollama)
%% Session A: Conversational Filing Flow
Note over User, AI: Pathway A: Conversational Case Filing (Standard REST Flow)
User->>FE: Submits query / voice input in Indic language
FE->>BE: POST /api/vakil-friend/chat/{sessionId} (English or Indic)
activate BE
BE->>BHA: Speech-To-Text / Translate input query to English
BHA-->>BE: Returns English translation
BE->>GPT: POST /context (English query RAG lookup)
activate GPT
GPT-->>BE: Returns top-3 FAISS semantic document chunks
deactivate GPT
BE->>AI: Chat completion request (System Prompt + Legal Context + English query)
activate AI
AI-->>BE: Returns structured legal response (English)
deactivate AI
BE->>BHA: Translate response back to original Indic language
BHA-->>BE: Returns translated response
BE->>DB: Log encrypted interaction to Case Diary (SHA-256 protected)
BE-->>FE: Return localized response & readyToFile heuristic flag
deactivate BE
FE-->>User: Renders text response & speaks via TTS / lip-syncs Avatar
%% Session B: Deep Legal Research Flow
Note over User, AI: Pathway B: Deep Legal Research (SSE Streaming Flow)
User->>FE: Submits complex legal issue under Avatar page
FE->>NLP: EventSource: POST /research/deep (Starts SSE Connection)
activate NLP
NLP->>AI: Stage 1: Detect domain & query complexity
AI-->>NLP: Domain classification details
NLP-->>FE: SSE: stage [understanding]
NLP->>AI: Stage 2: Decompile into sub-questions & route
AI-->>NLP: Sub-questions returned
NLP-->>FE: SSE: sub_questions
NLP->>AI: Stage 3: Query Indian Kanoon API for search criteria
AI-->>NLP: Search results / Context returned
NLP-->>FE: SSE: kanoon_results (Top citations)
NLP->>AI: Stage 4: Run Parallel Research (Groq / Gemini)
AI-->>NLP: Streaming sub-answers / reasoning
NLP-->>FE: SSE: reasoning (Real-time token chunks)
NLP->>AI: Stage 5: Verdict & synthesis (Verify citations + Convert to Hinglish)
AI-->>NLP: Final Hinglish dialogue + validated markdown
NLP-->>FE: SSE: final (Markdown, Hinglish audio dialogue, citations)
NLP-->>FE: SSE: done
deactivate NLP
FE-->>User: Streams markdown verdict to screen & speaks Hinglish speech via Avatar
This project implements a robust API versioning strategy starting from v1.
- The stable release of our APIs is accessible under the
/api/v1/prefix. - A global
WebMvcConfigurerautomatically prefixes all backend controllers. - Any future breaking changes will be introduced under a new version (e.g.,
/api/v2/), ensuring backward compatibility for existing client applications.
For complete setup instructions, environment configuration, and Docker deployment details, refer to the Detailed Setup Guide.
At a high level, the platform consists of three services that need to run concurrently:
| Service | Directory | Command |
|---|---|---|
| Frontend | frontend/nyaysetu-frontend/ |
npm install && npm run dev |
| Backend | backend/nyaysetu-backend/ |
mvn spring-boot:run |
| NLP Orchestrator | nlp-orchestrator/ |
uvicorn main:app --reload |
Prerequisites: Node.js >= 20, Java 17, Maven 3.9+, PostgreSQL 15+, Python 3.12+
For environment variables, copy .env.example to .env and fill in your values. A full reference of all required variables is documented in the Setup Guide.
| Document | Description |
|---|---|
| Setup Guide | Full database setup, environment variables, and Docker configuration |
| Architecture Overview | System design, component diagrams, and data flow |
| AI Integration Guide | Groq API and NLP orchestrator technical deep-dive |
| API Documentation | All REST endpoints with request and response specifications |
| Contributing Guidelines | Branching strategy, commit conventions, and PR workflow |
This project is part of GSSoC (GirlScript Summer of Code) 2026. Contributions are welcome from everyone, regardless of experience level.
Before opening a Pull Request:
| Requirement | Details |
|---|---|
| Read the guidelines | Review CONTRIBUTING.md for our branching and commit conventions |
| Link an issue | Every PR must reference the issue it closes (Closes #123) |
| Include visuals | UI-affecting PRs must include screenshots or a screen recording |
| Sync with main | Rebase or merge main into your branch before requesting review |
| Pass all checks | CI checks for lint, tests, and build must all pass |
Browse open issues and filter by good first issue to get started.
- Ensure Node.js version is
>= 20 - Delete
node_modulesand reinstall dependencies:
rm -rf node_modules
npm install- Verify PostgreSQL service is running
- Check database credentials in
.env - Ensure database and user are created properly
Verify Java version:
java -versionEnsure Java 17 is installed and configured correctly.
- Copy
.env.exampleto.env - Fill all required variables before starting services
- Ensure PostgreSQL is running
- Verify Maven installation:
mvn -version- Check all dependencies are installed properly
Stop conflicting processes or use different ports.
Example:
npx kill-port 3000This chart updates automatically as new contributors merge pull requests. Want to see your avatar here? Pick up an issue and start contributing.
This project is licensed under the MIT License. See the LICENSE file for the full license text and guidelines on bulk-adding license headers to all source files in the repository.
Built with purpose for a more accessible Indian Judiciary.
Nyay Setu — न्याय हर किसी का अधिकार है।