Jun 2026
CampyWin ML (MLOps)

Code source privé
À propos de ce projet
An MLOps platform around a price-prediction model trained on 961 Decathlon Tunisia products. FastAPI serves the model from the MLflow registry, Prefect runs six pipelines, metrics flow through the ELK stack, and a Next.js dashboard monitors the eight containers.
Points clés
- Gradient Boosting tuned with GridSearchCV, 29 registered model versions
- Prefect pipelines (train, evaluate, experiment…) triggered from the dashboard
- Docker Compose stack: FastAPI, MLflow, Prefect, Elasticsearch, Logstash, Kibana
Technologies utilisées
Pythonscikit-learnMLflowPrefectFastAPIElasticsearchKibanaNext.jsDocker


