Tech Stack Overview

This page summarizes MeetEasy's technology choices for business stakeholders and technical evaluators. For implementation details, see the Developer Guide.

At a Glance

Layer	Technologies
Backend API	Python 3.10+, FastAPI
Database	MongoDB + Beanie ODM
Messaging & async	FastStream + Redis
Real-time	Socket.IO
Frontend	Vue 3, TypeScript, Pinia, Vite
Mobile UI	Vant UI (H5), UniApp (WeChat)
Admin UI	Ant Design Vue
Low-code	AMIS (forms), VisuSpace DSL (microsites)
Infrastructure	MongoDB, Redis, Nginx

Backend

• FastAPI — High-performance async API framework with automatic OpenAPI documentation.
• Beanie — Async MongoDB ODM for document-based data models.
• FastStream — Message-driven async task processing over Redis.
• Socket.IO — Bidirectional real-time communication for notifications and live updates.

The backend follows a layered architecture: routers → services → CRUD → models, with dependency injection for auth, tenant context, and database connections.

Frontend

All frontend applications share a pnpm monorepo workspace:

App	Stack	Target
Admin	Ant Design Vue + TailwindCSS	Desktop browser
Console	Vant UI + TailwindCSS	Mobile H5
MeetApp	Vant UI + TailwindCSS	Mobile H5
VisuSpace	Custom editor + TailwindCSS	Desktop browser
WeApp / WeConsole	UniApp + Vue 3	WeChat mini programs

• WebAPI SDK (src/frontend/webapi) — Shared TypeScript API client used by all frontends.
• vue-i18n — Bilingual support (zh-CN / en-US) across all surfaces.
• Pinia — Centralized state management.

Infrastructure

• MongoDB — Primary document store for conferences, registrations, users, and analytics.
• Redis — Message queue, caching, and session storage.
• Nginx — Reverse proxy, static asset serving, and SSL termination.
• Docker Compose — Local development and optional production deployment.

AI & Voice Processing

MeetEasy provides a flexible hybrid architecture for AI and voice processing, supporting both local isolated inference (via AI Compute Worker) and external cloud SaaS APIs:

OCR (Optical Character Recognition)

• Local/Offline Inference: Powered by the PaddleOCR engine. To save memory and prevent API latency spikes, the models are loaded inside the isolated ai-worker process.
• Cloud API Support: Pre-integrated with Alibaba Qwen-VL and Baidu Cloud OCR services.
• Key Use Cases: Automatic text extraction from conference handbooks or schedule images to auto-populate agenda and speaker databases.

ASR (Automatic Speech Recognition)

• Local/Offline Inference: Built on Alibaba's FunASR framework (e.g., Paraformer) and OpenAI's Whisper architecture, providing high-precision Chinese/English speech-to-text. Runs inside the isolated ai-worker.
• Cloud API Support: Integrates Alibaba Cloud ASR, Baidu ASR, and other mainstream speech APIs.
• Key Use Cases: Speech-to-text input inside the MeetApp AI Chatbox to support natural voice commands and hands-free operations.

LLM & Page-Generation Agent

• Large Language Models: Supports tenant-configurable routing to DeepSeek, OpenAI, and other providers via the plugin system.
• VisuSpace AI: A specialized page-generation agent that takes natural language prompts and outputs standardized VisuSpace JSON DSL configurations.

WeChat Integration

• Platforms: WeChat Mini Program (WeApp for attendees, WeConsole for organizers).
• Key Features: WeChat login, Single Sign-On (SSO) ticket exchange, QR code scanning, and card sharing.

Why These Choices

Requirement	Technology choice
Flexible schema for diverse event data	MongoDB document model
Real-time attendee notifications	Socket.IO
Non-blocking background jobs	FastStream + Redis
Rapid mobile UI development	Vant UI + UniApp
Shared API across six apps	Monorepo + WebAPI SDK
Low-code registration forms	AMIS
Visual microsite builder	VisuSpace DSL

Deployment Model

MeetEasy supports self-hosted deployment via shell scripts, systemd services, and Nginx configuration. See Deployment for operational details.

Production deployment targets (domain, hosting provider) are [TBD].