Architecture Deep Dive
This document provides a detailed look at ProAI's system architecture, components, and how they interact.
System Architecture
ProAI follows a microservices architecture deployed on Kubernetes. The platform is divided into four primary layers:
1. Client Layer
| Component | Description |
|---|---|
| Web Designer | React-based visual pipeline canvas |
| Notebooks | Jupyter-compatible notebook environment |
| CLI | Command-line interface for automation |
| REST API | RESTful API for programmatic access |
| Dashboard | Monitoring and management console |
2. Service Layer
| Service | Responsibility |
|---|---|
| Project Manager | Project CRUD, version control, environments |
| DAG Engine | Dependency resolution, execution planning |
| Scheduler | Cron-based and event-driven scheduling |
| Auth Service | Authentication, authorization, SSO |
| Metadata Service | Schema registry, lineage tracking |
3. Execution Layer
| Component | Description |
|---|---|
| Connector Framework | Pluggable source/destination adapters |
| Transform Engine | SQL, Python, and Spark transformations |
| Agentic Engine | AI-powered workflow optimization |
| Streaming Runtime | Real-time event processing |
4. Platform Layer
| Component | Description |
|---|---|
| IAM | Identity and access management |
| Audit Logging | Compliance and activity tracking |
| Metrics | Performance and usage statistics |
| Alerting | Configurable alert rules and notifications |
Data Flow
Source Systems → Connectors → Transform Engine → Destinations
↑
Agentic Engine
(auto-optimization)
↓
Monitoring & Alerts
Deployment Model
ProAI supports three deployment models:
- ProAI Cloud — Fully managed SaaS
- Self-Hosted — Deploy on your own infrastructure
- Hybrid — Control plane in SaaS, data plane on-premises