← All Challenges

100DaysOfMLOps

A continuous learning journey to master Machine Learning Operations (MLOps), covering model versioning, training pipelines, deployment, and monitoring.

🔥
60 Days Current Streak
🎯
60% Completion
60/100 Total Days

The #100DaysOfMLOps challenge is a dedicated commitment to bridging the gap between machine learning model development and operational production. Over 100 days, I am learning and implementing tools to automate, monitor, and scale ML pipelines.

🎯 Learning Objectives

  • Data & Model Versioning: Tracking dataset versions and model artifacts using tools like DVC and MLflow.
  • Pipeline Automation: Building automated pipelines for data prep, training, and validation.
  • Model Deployment: Packaging models as microservices (FastAPI/Docker) and deploying to Kubernetes.
  • Monitoring & Observability: Tracking data drift, model performance decay, and latency in production.

🛠️ Core Topics Explored

  • Versioning: DVC (Data Version Control), Git LFS.
  • Tracking & Registry: MLflow, Weights & Biases.
  • Containerization & Orchestration: Docker, Kubernetes, Kubeflow, Prefect.
  • Model Serving: FastAPI, Seldon Core, Triton.
  • CI/CD for ML: GitHub Actions for automated model testing and build.