🚀 AI Software Engineering Team

An autonomous AI-powered software engineering platform that reviews GitHub repositories using 5 specialized AI agents, persistent RAG memory, and background job processing.

---

🌐 Live Demo

Frontend

https://ai-software-engineering-team.vercel.app

Backend API

https://ai-software-engineering-api.onrender.com/api/health

---

📌 Overview

AI Software Engineering Team automatically reviews GitHub repositories using multiple AI agents that work in parallel.

Instead of sending the entire repository to a single LLM prompt, the platform simulates a real software engineering team where each AI agent specializes in a different responsibility.

The system also includes persistent Retrieval-Augmented Generation (RAG) using ChromaDB, allowing previous findings to improve future repository reviews.

---

✨ Features

• 🔍 GitHub Repository Analysis
• 🤖 5 Parallel AI Agents
• 🧠 Persistent RAG Memory (ChromaDB)
• ⚡ BullMQ Background Processing
• 🔴 Redis Caching
• 📡 Real-Time Progress Updates
• 📊 Executive Summary Generation
• 📄 PDF Report Export
• 🗄 PostgreSQL + Prisma Storage
• 🔐 Authentication Ready
• 📈 Scan History Dashboard
• 🌍 Production Deployment

---

Backend Engineering Highlights

• Modular Node.js + Express architecture
• Background job processing using BullMQ + Redis
• Queue retry strategy and worker isolation
• PostgreSQL with Prisma ORM
• WebSocket communication using Socket.IO
• JWT Authentication
• RESTful API design
• Persistent vector storage with ChromaDB
• Event-driven architecture
• Production deployment across Render, Railway and Vercel

🧠 AI Agent System

The platform uses five specialized AI agents running simultaneously.

Agent	Responsibility
🏗 Architecture Agent	Software architecture review
🔒 Security Agent	Security vulnerability detection
💻 Code Review Agent	Code quality analysis
📦 Dependency Agent	Dependency inspection
⚡ Performance Agent	Performance bottleneck detection

After all agents complete their analysis, an AI Orchestrator combines their findings into a single engineering report.

---

🧠 Retrieval-Augmented Generation (RAG)

Unlike traditional AI code reviewers, this platform continuously learns from previous repository scans.

Workflow

Repository Scan
        │
        ▼
Generate Embeddings
        │
        ▼
Store Findings in ChromaDB
        │
        ▼
Future Repository Scan
        │
        ▼
Retrieve Similar Historical Findings
        │
        ▼
Provide Context to AI Agents
        │
        ▼
Generate Better Recommendations

Each successful scan expands the project's engineering knowledge base.

---

🏗 System Architecture

                        GitHub Repository
                               │
                               ▼
                   Repository Intelligence Engine
                               │
                               ▼
                      BullMQ Background Queue
                               │
                               ▼
                 ┌────────────────────────────┐
                 │ Parallel AI Agent System   │
                 └────────────────────────────┘
                     │    │    │    │    │
                     ▼    ▼    ▼    ▼    ▼
               Architecture Security Code Review
                 Performance Dependency
                     │
                     ▼
                AI Orchestrator
                     │
                     ▼
            Retrieval-Augmented Generation
                  (ChromaDB Memory)
                     │
                     ▼
             PostgreSQL + Prisma
                     │
                     ▼
          Socket.IO Real-Time Updates
                     │
                     ▼
               React Dashboard

---

⚙ Production Deployment

Service	Platform
Frontend	Vercel
Backend API	Render
AI Worker	Railway
Vector Database	Render (ChromaDB)
PostgreSQL	Prisma Postgres
Queue	BullMQ
Cache	Redis

---

🚀 Engineering Workflow

User submits GitHub URL
            │
            ▼
Repository Extraction
            │
            ▼
BullMQ Queue
            │
            ▼
5 Parallel AI Agents
            │
            ▼
RAG Context Retrieval
            │
            ▼
AI Orchestrator
            │
            ▼
Save Report
            │
            ▼
Update RAG Memory
            │
            ▼
Dashboard

---

💻 Tech Stack

Frontend

• React
• React Router
• Zustand
• Axios
• Vite

Backend

• Node.js
• Express.js
• Prisma ORM
• PostgreSQL
• BullMQ
• Redis
• Socket.IO

AI Layer

• GPT-4o
• Multi-Agent Architecture
• Repository Intelligence Engine
• ChromaDB
• Transformers.js Embeddings
• Retrieval-Augmented Generation (RAG)

---

📂 Project Structure

AI-Software-Engineering-Team

backend
│
├── api
├── worker
├── agents
├── orchestrator
├── rag
├── queue
├── prisma
├── github
├── reports

frontend
│
├── components
├── pages
├── hooks
├── api
├── store
├── utils

---

📈 Current Capabilities

✅ Repository Intelligence

✅ Multi-Agent Analysis

✅ Historical AI Memory

✅ Semantic Search

✅ Parallel Processing

✅ Background Jobs

✅ Executive Reports

✅ Real-Time Dashboard

✅ Persistent Scan History

---

🔮 Roadmap

• Pull Request Reviews
• CI/CD Integration
• Team Collaboration
• Multi-Language Analysis
• Custom Rule Engine
• Repository Comparison
• AI Fix Suggestions
• GitHub App Integration
• Automated Code Refactoring

---

⚙ Environment Variables

DATABASE_URL=

GITHUB_TOKEN=

OPENAI_API_KEY=

REDIS_HOST=

REDIS_PORT=

REDIS_USERNAME=

REDIS_PASSWORD=

UPSTASH_REDIS_URL=

UPSTASH_REDIS_TOKEN=

CHROMA_URL=

---

🚀 Installation

Backend

cd backend

npm install

npm run dev

Frontend

cd frontend

npm install

npm run dev

---

📹 Demo

https://www.loom.com/share/fa38986215b84c61a82d1f7e66a601b5

---

👨‍💻 Author

Ankit Tripathi

Full Stack Developer | AI Engineer

GitHub

https://github.com/ankittrip

LinkedIn

https://linkedin.com/in/ankittripathi-dev

---

⭐ Support

If you found this project helpful, please consider giving it a ⭐ on GitHub.

It motivates further development and helps others discover the project.