🎨 Beri ML - AI Team Development IDE

Beri ML is a revolutionary IDE where AI developers work together as a team to build applications autonomously. Watch Cheri-ML (Backend), Beri-ML (Frontend), and Sheri-ML (Full Stack) collaborate in real-time with human oversight.

🌟 Vision

Imagine VS Code where 3 AI engineers work alongside you:

Cheri-ML 🧠 - ML/Backend Engineer (APIs, databases, ML models)
Beri-ML 🎨 - UX/Frontend Engineer (UI/UX, React, design systems)
Sheri-ML ⚡ - Full Stack Developer (Integration, deployment, DevOps)

Each has 3 junior developers:

Researcher - Reads docs, analyzes patterns, suggests architectures
Developer - Writes production code
Tester - Creates tests, reviews code, ensures quality

🏗️ Architecture

┌─────────────────────────────────────────────────────────────────┐
│                    Beri ML IDE (VS Code Fork)                   │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  👨‍💻 Cheri-ML          🎨 Beri-ML           👩‍💻 Sheri-ML          │
│  Backend Engineer    Frontend Engineer   Full Stack Dev       │
│                                                                 │
│  ┌─────────────┐     ┌─────────────┐     ┌─────────────┐     │
│  │ Junior 1 📖 │     │ Junior 1 📖 │     │ Junior 1 📖 │     │
│  │ Research    │     │ Research    │     │ Research    │     │
│  ├─────────────┤     ├─────────────┤     ├─────────────┤     │
│  │ Junior 2 💻 │     │ Junior 2 💻 │     │ Junior 2 💻 │     │
│  │ Coding      │     │ Coding      │     │ Coding      │     │
│  ├─────────────┤     ├─────────────┤     ├─────────────┤     │
│  │ Junior 3 ✅ │     │ Junior 3 ✅ │     │ Junior 3 ✅ │     │
│  │ Testing     │     │ Testing     │     │ Testing     │     │
│  └─────────────┘     └─────────────┘     └─────────────┘     │
│                                                                 │
│  Status: Working on auth    Designing login    Integrating    │
│  Lines: 1,247/5,000         Lines: 892/5,000   Lines: 1,456   │
└─────────────────────────────────────────────────────────────────┘

✨ Key Features

1. Multi-Agent Collaboration

3 senior AI engineers, each with 3 junior assistants
Real-time collaboration with human oversight
Autonomous task distribution via Kanban board

2. Live Team Chat

Agents discuss features, bugs, and architecture
Scheduled "Office Hours" every 2 hours
Natural conversation powered by Gemini 3.0

3. 9-Pane Editor

Each junior dev has their own editor pane
Watch them research, code, and test simultaneously
5,000 line focus limit per agent for quality

4. Human Approval Workflow

All AI-generated code requires approval
Side-by-side diff view
Live simulator preview for mobile apps

5. Integrated Build Pipeline

AWS EC2 Mac fleet for iOS/macOS builds
Live iPhone simulator streaming in browser
One-click deployment to TestFlight

6. Cost-Optimized AI

Google Vertex AI for 60-80% cost reduction
Ollama integration for local inference
Smart model routing (local vs cloud)

🛠️ Tech Stack

IDE Core

Base: VS Code OSS (Electron + Monaco Editor)
Language: TypeScript
UI: React + TailwindCSS
State: Zustand
Real-time: Socket.IO

AI Models

Cheri-ML: Custom fine-tuned Deepseek 1.3B
Google Gemini: 3.0 Flash & Pro (research, voice, multimodal)
OpenAI: GPT-4 Turbo (frontend coding)
Anthropic: Claude 3.5 Sonnet (code review)
Ollama: Local models (Mixtral, Llama3, Phi3)

Infrastructure

Build Servers: AWS EC2 Mac2-m2 (M2 chip)
API Gateway: Cloudflare Workers
Database: Cloudflare D1 + Supabase
Storage: Cloudflare R2
Queue: BullMQ + Redis

📦 Project Structure

beri-ml/
├── ide/                          # VS Code fork
│   ├── extensions/
│   │   └── beri-ml-team/        # Main extension
│   │       ├── agent-orchestrator.ts
│   │       ├── agents/
│   │       │   ├── cheri-ml.ts
│   │       │   ├── beri-ml.ts
│   │       │   └── sheri-ml.ts
│   │       ├── views/
│   │       │   ├── team-view.ts
│   │       │   ├── kanban-view.ts
│   │       │   └── chat-view.ts
│   │       └── models/
│   │           ├── model-router.ts
│   │           ├── vertex-ai.ts
│   │           ├── ollama.ts
│   │           └── cheri-ml-client.ts
│   └── package.json
│
├── backend/                      # API & Orchestration
│   ├── src/
│   │   ├── api/                 # REST API
│   │   ├── agents/              # Agent logic
│   │   ├── build/               # Build pipeline
│   │   └── models/              # AI model integrations
│   └── wrangler.toml
│
├── mac-setup/                    # AWS EC2 Mac configuration
│   ├── setup-mac-builder.sh    # Initial setup script
│   ├── install-xcode.sh        # Xcode installation
│   ├── configure-simulators.sh # Simulator setup
│   ├── hadp-agent/             # Build agent daemon
│   └── README.md
│
├── docs/                         # Documentation
│   ├── ARCHITECTURE.md
│   ├── API.md
│   ├── AGENTS.md
│   ├── SETUP_AWS_MAC.md
│   ├── SETUP_APPLE_DEVELOPER.md
│   └── GPU_AGENT_PROMPT.md
│
├── scripts/                      # Utility scripts
│   ├── deploy.sh
│   ├── test.sh
│   └── cost-report.sh
│
└── README.md                     # This file

🚀 Quick Start

Prerequisites

Node.js 20+
Git
AWS Account (for EC2 Mac)
Apple Developer Account
Google Cloud Account (for Vertex AI)
API Keys: OpenAI, Anthropic, Cheri-ML

1. Clone Repository

git clone https://github.com/Hey-Salad/beri-ml.git
cd beri-ml

2. Install Dependencies

# Install IDE dependencies
cd ide
npm install

# Install backend dependencies
cd ../backend
npm install

3. Configure Environment

# Copy example env file
cp .env.example .env

# Add your API keys
nano .env

4. Start Development

# Terminal 1: Start backend
cd backend
npm run dev

# Terminal 2: Start IDE
cd ide
npm run watch

# Terminal 3: Launch Electron
cd ide
npm run electron

🏗️ Setup Tracks (Parallel Execution)

Track 1: AWS EC2 Mac Setup ☁️

Assignee: Claude Agent Duration: 2-3 hours Steps:

Launch EC2 Mac2-m2 instance
Run mac-setup/setup-mac-builder.sh
Install Xcode and simulators
Configure build agent
Test with simple iOS build

📖 Guide: docs/SETUP_AWS_MAC.md

Track 2: Beri ML IDE Core 💻

Assignee: GPU Agent (You or another AI) Duration: 1 week Steps:

Fork VS Code OSS
Build agent orchestrator system
Implement 3-agent architecture
Create 9-pane editor UI
Add Kanban board and chat interface

📖 Prompt: docs/GPU_AGENT_PROMPT.md

Track 3: Apple Developer Setup 🍎

Assignee: Human (You) Duration: 1 hour Required:

📖 Guide: docs/SETUP_APPLE_DEVELOPER.md

Track 4: Integration & Testing 🧪

Assignee: All (Collaborative) Duration: 2-3 days Steps:

Connect IDE to EC2 Mac build servers
Test agent code generation → build → simulator
Deploy test app to TestFlight
Optimize and polish

💰 Cost Estimate

Component	Monthly Cost
AWS EC2 Mac2-m2 (24/7)	$1,800
Google Vertex AI	$1,500-2,000
Cloudflare (Workers + D1 + R2)	$50
Redis (Upstash)	$50
Total	~$3,400-3,900/month

With Ollama optimization: ~$2,000-2,500/month

Compare to:

3 developers × $100k/year = $25k/month
Savings: 90%+ 🎉

📊 Model Selection Strategy

Task	Agent	Model	Provider	Cost/1M	Reason
Backend Research	Cheri Junior 1	Gemini Flash	Vertex AI	$0.075	Fast & cheap
Backend Coding	Cheri Junior 2	Cheri-ML 1.3B	HeySalad	$0.10	Custom trained
Backend Testing	Cheri Junior 3	Claude 3.5	Vertex AI	$3.00	Thorough
Frontend Research	Beri Junior 1	Gemini Pro	Vertex AI	$1.25	Multimodal
Frontend Coding	Beri Junior 2	GPT-4 Turbo	Vertex AI	$10.00	Best at React
Frontend Testing	Beri Junior 3	Claude 3.5	Vertex AI	$3.00	UI testing
Integration Research	Sheri Junior 1	Gemini Flash	Vertex AI	$0.075	Fast docs
Integration Coding	Sheri Junior 2	GPT-4 Turbo	Vertex AI	$10.00	Versatile
DevOps/Deploy	Sheri Junior 3	Claude Opus	Direct	$15.00	System-level

🎯 Roadmap

Phase 1: Foundation (Week 1-2) ✅

Phase 2: Core IDE (Week 3-4)

Phase 3: AI Integration (Week 5-6)

Phase 4: Build Pipeline (Week 7-8)

Phase 5: Polish & Launch (Week 9-10)

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

How to Contribute

Fork the repository
Create a feature branch (git checkout -b feature/amazing-feature)
Commit your changes (git commit -m 'Add amazing feature')
Push to the branch (git push origin feature/amazing-feature)
Open a Pull Request

📄 License

MIT License - see LICENSE file for details.

🙏 Acknowledgments

VS Code Team - For the amazing open source IDE
OpenAI - For GPT models
Google - For Gemini and Vertex AI
Anthropic - For Claude models
HeySalad Team - For Cheri-ML and infrastructure

📞 Support

Website: heysalad.com
Email: support@heysalad.com
Discord: Join our community
Issues: GitHub Issues

🌟 Star History

Built with ❤️ by HeySalad Inc. • San Francisco, CA

Name		Name	Last commit message	Last commit date
Latest commit History 10 Commits
.github/workflows		.github/workflows
cloudflare		cloudflare
docs		docs
mac-setup		mac-setup
scripts		scripts
.env.example		.env.example
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
WORK_TRACKS.md		WORK_TRACKS.md
package.json		package.json

Folders and files

Latest commit

History

Repository files navigation

🎨 Beri ML - AI Team Development IDE

🌟 Vision

🏗️ Architecture

✨ Key Features

1. Multi-Agent Collaboration

2. Live Team Chat

3. 9-Pane Editor

4. Human Approval Workflow

5. Integrated Build Pipeline

6. Cost-Optimized AI

🛠️ Tech Stack

IDE Core

AI Models

Infrastructure

📦 Project Structure

🚀 Quick Start

Prerequisites

1. Clone Repository

2. Install Dependencies

3. Configure Environment

4. Start Development

🏗️ Setup Tracks (Parallel Execution)

Track 1: AWS EC2 Mac Setup ☁️

Track 2: Beri ML IDE Core 💻

Track 3: Apple Developer Setup 🍎

Track 4: Integration & Testing 🧪

💰 Cost Estimate

📊 Model Selection Strategy

🎯 Roadmap

Phase 1: Foundation (Week 1-2) ✅

Phase 2: Core IDE (Week 3-4)

Phase 3: AI Integration (Week 5-6)

Phase 4: Build Pipeline (Week 7-8)

Phase 5: Polish & Launch (Week 9-10)

🤝 Contributing

How to Contribute

📄 License

🙏 Acknowledgments

📞 Support

🌟 Star History

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages