Case study
Smart Survey — container damage detection
AI-assisted container survey: detects damage from photos for yard inspectors.
FIG.1AI
- Role
- Sole architect and developer
- Stack
- PythonGeminiOpenAIAngularPostgreSQLDocker
- Result
- Fewer moving parts to operate, and a first-pass damage call on every photo.
Problem
Damage surveys were slow and inconsistent between inspectors.
What I owned
Sole architect and developer
What I built
- Designed and built the AI worker and the front end.
- Removed four stateful dependencies in favor of a lean PostgreSQL-only stack.
System shape
Worker running GenAI vision (OpenAI/Gemini) with an Angular front end. Simplified from Keycloak, MinIO, Neo4j, and RabbitMQ down to PostgreSQL with REST and WebSocket.
What changed
Fewer moving parts to operate, and a first-pass damage call on every photo.