Self-hosted Architecture

This documentation is automatically updated from our Platform Roadmap Miro board. For the latest architectural diagrams and detailed workflows, https://miro.com/app/board/uXjVIxf4RDw=/

Overview

Our self-hosted architecture is designed to provide enterprise-grade security, scalability, and reliability for agentic AI systems. The architecture is built around a Service Fabric model that supports multi-tenancy, cross-team collaboration, and comprehensive lifecycle management.

Architecture Components

1. Cross-team Workflow

Supports collaboration across engineering teams with clear coordination and structure:

• Standardized Interfaces: Shared APIs and messaging protocols streamline communication and integration.

• Managed Dependencies: Teams align on service contracts and reuse common infrastructure through defined collaboration patterns.

2. Team Boundaries

Maintains performance, security, and organizational clarity through isolation and ownership:

• Isolated Resources & Networks: Dedicated compute, storage, and access controls per team.

• Service Accountability: Clear service ownership and inter-team APIs ensure modular, maintainable systems.

3. Tenant Management

Tenant Onboarding	Tenant Boundaries
Enables secure, consistent, and scalable setup for new tenants: Automated Setup: Infrastructure, security, and monitoring provisioned via code. Tailored Configuration: Dynamic resource allocation and tenant-specific settings.	Guarantees strong tenant isolation across system layers: Multi-level Isolation: Separate network, compute, and encrypted data boundaries. Scoped Access: Role-based controls and full audit logging per tenant.

4. Engineering Lifecycle

Covers the full pipeline from development to production monitoring:

• End-to-end Automation: Includes dev environments, testing, and CI/CD pipelines.

• Reliable Delivery: Staged releases, rollbacks, and observability ensure quality in production.

5. Application Team Workflow

Optimized for productivity, reliability, and resilience:

• Development Best Practices: Standardized coding, reviews, and deployment strategies.

• Operational Excellence: Auto-scaling, efficient resource use, and responsive incident handling.

---

Infrastructure Details

Infrastructure Information

DuploCloud Infrastructure

Our infrastructure is managed through DuploCloud, providing automated cloud operations:

GCP Projects and Clusters

Production Environment:

• GCP Project: sullyai-prod

• Cluster: duploinfra-prod

• Region: us-west1

Non-Production Environment:

• GCP Project: sullyai-nonprod

• Cluster: duploinfra-nonprod

• Region: us-west1

Production V2 Environment:

• GCP Project: sullyai-prod-v2

• Cluster: gke-global-primary

• Region: us-central1-b

DuploCloud Namespaces

Development:

• Namespace: DEV01

• Purpose: Development and testing workloads

• Access Pattern: http://infra-proxy/api/v1/SERVICE_A/*

Staging:

• Namespace: STAGE

• Purpose: Pre-production validation

• Access Pattern: http://infra-proxy/api/v1/SERVICE_A/*

Production:

• Namespace: PROD

• Purpose: Production workloads

• Access Pattern: http://infra-proxy/api/v1/SERVICE_A/*

Service Routing

Services are accessed through a standardized routing pattern:

• Development: SERVICE_A (Namespace dev) /api/v1/

• Staging: SERVICE_A (Namespace stag) /api/v1/

• Production: SERVICE_A (Namespace prod) /api/v1/