Project Experience

AI Application Generation Platform (2025.11 - 2026.01)

GitHub: https://github.com/Zonest798/ai-code-monther

Live Demo: http://www.zonest.top

Background

Inspired by Meituan's NoCode low-code platform, built with Spring Boot 3 + LangChain4j + LangGraph4j. Users input natural language requirements, and the AI agent autonomously performs material collection, code generation, code quality validation, and project building, culminating in one-click deployment to a live web application.

Key Contributions

1. Multi-vendor LLM Integration: Integrated DeepSeek, Tongyi Qianwen and other LLMs via LangChain4j, using AI Service declarative programming for rapid model invocation to power automatic code generation;
2. AI Multi-turn Conversation Persistence: Used Redis for conversation history storage with memoryId to isolate user sessions, supporting multi-turn dialogue for iterative frontend project refinement;
3. AI Streaming Output Visualization: Custom TokenStream processor for real-time parsing and rendering of AI agent internal tool call processes, transforming the black-box execution into a user-friendly visual display;
4. Engineering Standardization: Adopted layered architecture with multiple design patterns for business logic decoupling; unified LLM invocation and Redis session storage utilities; provided complete API and deployment documentation with standardized code comments;预留 multi-module parallel development extension interfaces for future team collaboration and rapid onboarding.