All case studies
Master of SWE & AI · Cloud Computing (Assessment 3)·2026
Self-hosted BI with Apache Superset
Deploying open-source BI end-to-end on two clouds — and learning where deployments actually break.
DataAutomation
Problem
Deploy an open-source BI platform end-to-end on cloud infrastructure and prove it’s production-usable — more than a login screen — with real data, dashboards, and role-based access.
Approach
- Provisioned on Azure: Resource Group, VNet + subnet, Network Security Group, Ubuntu VM, Docker Compose (Superset + PostgreSQL + Redis).
- Locked the NSG down: opened port 8088 intentionally, kept SSH on 22 restricted to my IP, denied everything else inbound.
- Proved it was real: uploaded CSVs, built a working dashboard, configured Admin / Alpha / Gamma RBAC roles.
- Repeated the stack on AWS EC2 and worked through the differences (dnf vs apt, Docker Compose not bundled, 8088 over an SSH tunnel).
Stack
Apache SupersetAzureAWS EC2Docker ComposePostgreSQLRedisNSG / networking
Outcome
- A working BI deployment with dashboards and RBAC, reproduced on two different clouds.
- Hard-won deployment-friction lessons: architecture diagrams show intent — deployment shows what you actually understand.