Backend · AI/ML · DevOps
AERS — AI Emergency Response System
AERS is an AI-driven emergency triage backend that classifies incoming cases, scores them using NEWS2 / qSOFA standards, and dispatches the nearest ambulance in real time.
Overview
AERS is an AI-driven emergency triage backend that classifies incoming cases, scores them using NEWS2 / qSOFA standards, and dispatches the nearest ambulance in real time.
Problem
Emergency response teams need fast, explainable triage decisions — and the software that supports them must be testable, reproducible and deployable on commodity infrastructure.
Solution
Built a FastAPI backend exposing prediction endpoints, containerized the entire stack with Docker, exposed predictions via a Streamlit dashboard, and added SHAP for explainability.
Technology
- Python
- FastAPI
- Streamlit
- Docker
- pytest
- WebSockets
Technical Highlights
- SHAP-based model explainability
- NEWS2 + qSOFA scoring pipelines
- WebSocket updates for live triage
- Ambulance dispatch logic module
- 57-test pytest suite for the API
Results
- 57 tests passing across the API surface
- Containerized deployment via Docker
- Explainable predictions through SHAP