LexAI

AI-powered legal technology platform built to connect people with legal help, legal resources, and modern digital workflows across mobile, web, and backend services.

It combines a Flutter mobile experience, a React/Vite web workspace, and a FastAPI service layer with hybrid retrieval, WebSockets, and voice-enabled capabilities. The repository is laid out to support product development, experimentation, and future scaling in one place.

🚀 Live Deployments

• Web Application Workspace: https://lexai-3fd1a.web.app (Deployed via Firebase Hosting, featuring Google Sign-In and real-time legal consultation queries)
• Production RAG API Backend: https://core-api-584212158273.asia-south1.run.app (FastAPI containerized service running on Google Cloud Run with 4GB RAM to load dense FAISS indices & sentence-transformers under full CORS authorization)

Why this project stands out

This codebase is more than a demo app. It presents a full-stack product direction with a recruiter-friendly story: a real-world problem space, a multi-platform UI layer, a Python backend, AI-assisted legal interaction, and a research-backed retrieval pipeline.

Highlights:

• Multi-platform client strategy with Flutter for mobile and React/Vite for web experimentation.
• FastAPI backend designed for production-style API delivery, real-time communication, and service orchestration.
• Hybrid legal search layer that combines sparse, dense, and graph-based retrieval ideas.
• Voice-oriented workflows that support transcription and conversational legal assistance.
• Research and documentation folders that show product thinking, evaluation, and technical depth.

Core product capabilities

The current project direction supports several practical legal-tech workflows:

• AI legal chat for guided legal support and question answering.
• Lawyer discovery and client-to-lawyer communication flows.
• Document upload and review-oriented interactions.
• Voice capture and transcription pipelines for hands-free input.
• User and lawyer dashboards with localized mobile UX.
• Authentication, profile management, notifications, help, and account-related app flows.

Architecture

The repository is organized as a monorepo so each layer can evolve independently while still fitting into one product vision.

graph TD
    subgraph Clients [Client Layer]
        MobileApp[apps/mobile_app <br> Flutter Mobile]
        WebApp[apps/web_app <br> Web Playground]
    end

    subgraph Services [Service Layer]
        FastAPI[services/core_api <br> FastAPI Production API]
        Streamlit[services/core_api <br> Streamlit Research UI]
        HybridSearch[services/core_api/hybrid_search <br> Sparse + Dense + Graph]
    end

    subgraph Research [Research Layer]
        Benchmarks[research/benchmarks]
        Papers[research/research_paper]
    end

    MobileApp -->|HTTP / WebSockets| FastAPI
    WebApp -->|HTTP| FastAPI
    Streamlit -->|Internal Import / REST| HybridSearch
    FastAPI -->|Internal Import| HybridSearch

    HybridSearch -.->|Evaluation Data| Benchmarks

Loading

Additional flow view

flowchart LR
    User[User] --> Mobile[Flutter app]
    User --> Web[Web app]
    Mobile --> API[FastAPI backend]
    Web --> API
    API --> AI[Legal AI services]
    API --> Search[Hybrid retrieval]
    API --> Voice[Transcription / voice pipeline]
    AI --> Docs[Legal guidance and support]
    Search --> Docs

Loading

Repository structure

LexAI/
├── apps/
│   ├── mobile_app/        # Flutter mobile client
│   └── web_app/           # React/Vite web workspace
├── services/
│   └── core_api/          # FastAPI backend, hybrid search, and voice services
├── research/
│   ├── benchmarks/        # Experimental results and evaluation artifacts
│   ├── docs/              # Architecture and deployment documentation
│   ├── research_paper/    # Research-paper style documentation
│   └── wireframes/        # UI/UX drafts and layout exploration
└── README.md

Tech stack

• Flutter for the mobile client.
• React, Vite, and Axios for the web workspace.
• Python, FastAPI, and Streamlit for backend and research interfaces.
• Riverpod, GoRouter, and localization support in the Flutter app.
• WebSockets, voice recording, and audio playback packages for real-time interaction.
• Hybrid search components for retrieval-driven legal assistance.

Getting started

Backend

The backend lives in services/core_api and is managed with uv.

cd services/core_api
uv venv
.venv\Scripts\activate
uv pip install -r requirements.txt -r requirements_hybrid.txt
uv run app/main.py

The API is available at http://localhost:8001, with interactive docs at http://localhost:8001/docs.

To launch the research UI:

uv run streamlit run streamlit_app.py

Mobile app

The Flutter client lives in apps/mobile_app.

cd apps/mobile_app
flutter pub get
flutter gen-l10n
flutter run

Web workspace

The web app lives in apps/web_app.

cd apps/web_app
npm install
npm run dev

Testing

Run backend tests from the core API service:

cd services/core_api
uv run pytest tests/test_backend.py

Notes for reviewers

This repository is best read as a product-and-engineering portfolio piece. It shows:

• a clear problem domain,
• a layered full-stack architecture,
• UI clients for different platforms,
• backend services with AI and retrieval components,
• and supporting research artifacts that document the reasoning behind the system.

That makes it suitable for recruiters who want to see both implementation depth and product thinking in one repo.