MLOps for Energy Demand Forecasting

SeafaringIT

StageSur site4 à 6 moisDate limite : 23 févr. 2026
MLOps / DevOpsTime Series ForecastingData Science & Machine LearningAPIs & IntegrationsDevopsenergy analytics

Postuler

Description

  • Goals:
  • Build a production-ready ML system for energy demand forecasting
  • Optimize energy production, storage, and purchasing
  • Implement monitoring, retraining, and model improvement
  • Student roles: ML engineers, MLOps specialists, data engineers, backend engineers
  • Expected outcomes: MLOps pipeline with training, deployment, monitoring, and automated retraining for reliable predictions
  • Key features:
  • Dataset prep (consumption, production, weather) + feature engineering
  • Multiple time series models with comparison
  • MLflow for versioning/experiments
  • API deployment with batch jobs
  • Performance monitoring (latency, errors, concept drift)
  • Automated retraining
  • Operational guide, maintenance procedures
  • Cost and emissions impact analysis
  • Technologies: Python, time series libs (statsmodels, Prophet, LSTM), scikit-learn, MLflow, FastAPI, Docker, Docker Compose, CI/CD (GitHub Actions/GitLab CI), PostgreSQL, monitoring tools