Built by Dhairy Chauhan
Frontend (Vercel): https://system2-ml.vercel.app/
Backend (Render): https://system2ml-api.onrender.com
- Overview
- Features
- Tech Stack
- Architecture
- Getting Started
- API Endpoints
- Project Structure
- Deployment
- License
System2ML is a comprehensive, production-ready platform for designing, executing, and governing ML pipelines with AI-powered intelligence. It enforces user-defined constraints (cost, carbon, latency) throughout the entire ML lifecycle.
- 🔒 Constraint-Driven: Every pipeline validated against user constraints before execution
- 📊 Real Dataset Profiling: Accepts real data with auto type inference & PII detection
- ⚡ Live Monitoring: Real-time training with auto-stop capabilities
- 🎯 Feasibility Engine: Only generates pipelines that can actually run
- 🏢 Enterprise-Ready: OAuth, approvals, and complete audit trails
- 📁 Upload datasets (CSV, Parquet, JSON, images, text)
- 🔗 Connect to external sources (S3, GCS, Database, API)
- 🔍 Automatic profiling: data type, labels, size, class balance, PII detection
- Input - Review dataset summary
- Constraints - Set cost, carbon, latency limits
- Preferences - Configure objective, deployment, retraining
- Review - Validate constraints & generate candidates
- Results - Select feasible pipeline
- Pre-Training Gate - Final feasibility check
- Live Training - Real-time progress & constraint monitoring
- Results - Post-training metrics & model artifacts
- 🎨 Visual DAG editor with drag-and-drop
- 📦 10 node types (Source, Transform, Model, Sink, Monitor)
- 🔐 OAuth (Google, GitHub)
- ✅ Approval workflows
- 📝 Complete audit logs
- 💡 Failure memory with AI-suggested fixes
- 📈 Real-time drift detection
| Layer | Technology |
|---|---|
| Frontend | Next.js 16, TypeScript, Tailwind CSS, shadcn/ui |
| Backend | FastAPI, Python 3.11, Pydantic |
| Database | SQLite |
| Deployment | Vercel (Frontend), Render (Backend) |
| Authentication | OAuth (Google, GitHub) |
┌─────────────────────────────────────────────────────────────────┐
│ Frontend (Next.js 16) │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────────┐ │
│ │ Datasets │ │ Design │ │ Train │ │ Pipeline │ │
│ │ Pages │ │ Wizard │ │ Pages │ │ Designer │ │
│ └──────────┘ └──────────┘ └──────────┘ └──────────────┘ │
└───────────────────────────┬─────────────────────────────────────┘
│ REST API
┌───────────────────────────┴─────────────────────────────────────┐
│ Backend (FastAPI) │
│ ┌──────────────┐ ┌──────────────┐ ┌────────────────────┐ │
│ │ Validation │ │ Feasibility │ │ Training Engine │ │
│ │ API │ │ Policy │ │ │ │
│ └──────────────┘ └──────────────┘ └────────────────────┘ │
│ ┌──────────────┐ ┌──────────────┐ ┌────────────────────┐ │
│ │ Dataset │ │ Safety │ │ Eligibility │ │
│ │ Profiling │ │ Gate │ │ Matrix │ │
│ └──────────────┘ └──────────────┘ └────────────────────┘ │
└─────────────────────────────────────────────────────────────────┘
- Node.js 18+
- Python 3.10+
- pnpm or npm
# Clone the repository
git clone https://github.com/chauhand2463/System2ML.git
cd System2ML
# Install Python dependencies
pip install -e ".[all]"
# Install frontend dependencies
npm install
# or
pnpm install# Run both frontend and backend
npm run dev:all- Frontend: http://localhost:3000
- Backend: http://localhost:8000
| Method | Endpoint | Description |
|---|---|---|
| POST | /api/datasets/profile |
Profile a dataset |
| POST | /api/datasets/validate |
Validate against constraints |
| GET | /api/datasets |
List all datasets |
| Method | Endpoint | Description |
|---|---|---|
| POST | /api/validate |
Validate constraints |
| POST | /api/feasibility/policy |
Get feasibility policy |
| POST | /api/feasibility/generate |
Generate pipeline candidates |
| POST | /api/design/request |
Full design request |
| GET | /api/eligibility/matrix |
Model eligibility matrix |
| Method | Endpoint | Description |
|---|---|---|
| POST | /api/training/start |
Start training |
| GET | /api/training/{run_id} |
Get training status |
| POST | /api/training/{run_id}/stop |
Stop training |
| Method | Endpoint | Description |
|---|---|---|
| GET | /api/pipelines |
List pipelines |
| GET | /api/pipelines/{id} |
Get pipeline |
| POST | /api/pipelines/{id}/execute |
Execute pipeline |
| GET | /api/runs |
List runs |
system2ml/
├── app/ # Next.js App Router
│ ├── api/ # OAuth callbacks
│ ├── datasets/new/ # Dataset intake
│ ├── design/ # Design wizard
│ ├── train/ # Training pages
│ ├── pipelines/ # Pipeline management
│ ├── runs/ # Run history
│ ├── failures/ # Failure memory
│ ├── monitoring/ # Drift detection
│ ├── design-agent/ # AI proposals
│ ├── approvals/ # Approval workflow
│ ├── governance/ # Audit logs
│ └── login/ # Authentication
│
├── components/ # React components
│ ├── layout/ # Layout (sidebar, header)
│ ├── pipelines/ # Pipeline designer
│ ├── dashboard/ # Dashboard widgets
│ └── ui/ # shadcn/ui components
│
├── hooks/ # React hooks
│ ├── use-auth.tsx # Authentication
│ └── use-design.tsx # Design flow state
│
├── lib/ # Core libraries
│ ├── api.ts # API client
│ ├── types.ts # TypeScript types
│ ├── validation/ # Constraint validation
│ ├── feasibility/ # Policy engine
│ └── safety/ # Safety gate
│
├── agent/ # AI Agent
│ ├── planner.py # Design planner
│ └── rl_policy.py # RL policy
│
├── ui/ # FastAPI Backend
│ ├── api.py # All API routes
│ └── database.py # SQLite database
│
├── public/ # Static assets
├── pyproject.toml # Python config
├── package.json # Node config
└── README.md # This file
# Build and deploy
npm run buildEnvironment Variables:
| Key | Value |
|---|---|
NEXT_PUBLIC_API_URL |
https://system2ml-api.onrender.com |
| Setting | Value |
|---|---|
| Build Command | pip install -e ".[all]" |
| Start Command | uvicorn ui.api:app --host 0.0.0.0 --port $PORT |
| Python Version | 3.11 |
MIT License - Copyright (c) 2026 Chandan Kumar
⭐️ If you like this project, please give it a star on GitHub!
Built with ❤️ by Dhairy Chauhan
© 2026 System2ML - All rights reserved