Disaster Risk Observability Pipeline (DROP) is a data processing pipeline to predict the type, severity, and geo location of disasters events worldwide.
It leverages event-driven, and microservice architectures, and multilingual NER models to crawl, predict, and geocode historical, and realtime events at scale.
- Linux OS
- 8G RAM
- 8 CPUs
- Python ~3.10
- Docker Engine
Build pipeline steps
bash ./pipeline/build.sh
Single node
docker compose up
Multi node
docker stack deploy --detach=false -c compose.yaml -c compose.placement.yaml drop
For more info on Docker Swarm, read the Docker Documentation.