| Front-End |
Angular |
Web application portal for browsing, configuring, and invoking AI services |
| Back-End |
Python Flask |
Powers each AI service HTTP endpoint; orchestration logic; Azure component integration |
| Data Storage |
PostgreSQL |
Primary relational database for metadata, execution logs, and application state |
| Container Orchestration |
Azure Kubernetes Service (AKS) |
All AI services run in Docker containers, ensuring portability and scalability |
| Container Registry |
Azure Container Registry (ACR) |
Stores private Docker images for all AI services |
| Artifact Storage |
Azure Blob Storage |
Public Artifactory for service ZIPs; model artifacts; pipeline outputs |
| Secret Management |
Azure Key Vault + HashiCorp Vault |
All connection strings, API keys, credentials stored and retrieved at runtime |
| ML Experiment Tracking |
MLFlow |
Automatic experiment logging, comparison, and visualization for model training |
| GPU Compute (Cloud) |
RunPod, Denvr, Azure GPU VMs |
On-demand GPU nodes for training and heavy inference |
| Distributed Processing |
Ray |
Parallel task execution for large-scale video/data processing |
| Logging |
ELK Stack (Elasticsearch) |
Structured execution logs indexed for search and monitoring |
| IDE Integration |
Agile Workbench (VS Code plugin) |
Upload service ZIPs to Public Artifactory directly from VS Code |