Starship
An AI-assisted, self-service deployment platform that turned a manual, multi-day cloud rollout process into a guided workflow completed in hours — across seven cloud and on-premises targets.
The problem
Rolling out complex cloud-native workloads used to mean days of manual, specialist-driven work: back-and-forth requirement gathering over spreadsheets, hand-edited configuration, and error-prone environment setup that only a few engineers could perform. It didn't scale, and small mistakes were expensive.
What I built
Starship is a platform that automates the full path from "I need an environment" to "it's running and verified." A guided, self-service interface collects requirements, an AI layer proposes validated configuration, a human reviews and approves, and the platform provisions everything declaratively — consistently across every supported target.
How it works
- A guided interface collects deployment requirements in plain language instead of raw config files.
- An AI layer (LLM + retrieval over internal patterns) proposes environment-specific settings — compute, networking, storage, and orchestration values.
- Every AI suggestion passes through explicit human review before anything is provisioned — safety first.
- Approved parameters are translated into declarative infrastructure definitions and applied consistently.
- Post-deployment validation and health checks confirm the environment is correct before hand-off.
Impact
What used to take days of specialist effort now completes in hours, self-service, with a fraction of the configuration errors — and the same workflow works identically across every supported cloud and on-premises target.