MLOps Engineer BootCamp

Subtopics:

• What is Machine Learning?
• Types of ML (Supervised, Unsupervised, Reinforcement)
• Data preprocessing and EDA (overview only)
• Evaluation metrics (accuracy, precision, recall, AUC)
• Overfitting, underfitting, cross-validation
• Real-world ML use cases

Learning Objectives:

• Understand basic machine learning concepts and terminologies.
• Familiarize with ML types, metrics, and evaluation techniques.

Learning Outcomes:

• Able to differentiate between ML types and select appropriate metrics.
• Confident in understanding ML model structure and lifecycle.

Tools Used:

• Scikit-learn
• Pandas
• Jupyter Notebook

Subtopics:

• What is MLOps? Differences from DevOps
• MLOps lifecycle overview
• Tools and architecture in MLOps
• Azure ML & GCP Vertex AI overview
• Navigating Azure/GCP Console
• Creating and configuring ML Workspaces

Learning Objectives:

• Learn MLOps lifecycle and distinguish it from DevOps.
• Understand cloud platform components for MLOps.

Learning Outcomes:

• Able to set up ML workspace on Azure/GCP.
• Confident in navigating cloud consoles and MLOps tooling.

Tools Used:

• Azure ML
• Vertex AI
• Cloud Console

Subtopics:

• Data ingestion and preprocessing steps
• Building multi-step ML pipelines
• Using Azure ML Pipelines / Vertex AI Pipelines
• Reusability and modular pipeline design
• Logging & debugging pipelines

Learning Objectives:

• Design and build end-to-end ML workflows.
• Understand pipeline orchestration in cloud platforms.

Learning Outcomes:

• Able to build modular and reusable ML pipelines.
• Confident in debugging and scaling pipelines.

Tools Used:

• Azure ML Pipelines
• Vertex AI Pipelines

Subtopics:

• Experiment tracking (metrics, artifacts)
• Hyperparameter tuning
• Model registry (versioning, stage transitions)
• ML metadata management
• Integration with MLFlow (optional)

Learning Objectives:

• Track experiments and manage model lifecycle.
• Understand hyperparameter tuning and model versioning.

Learning Outcomes:

• Able to track, compare, and manage ML experiments.
• Able to register and deploy different model versions.

Tools Used:

• MLFlow
• Azure ML
• Vertex AI

Subtopics:

• Introduction to CI/CD for ML
• GitOps for ML pipelines
• Triggering ML workflows using GitHub Actions / Azure DevOps
• Automating retraining and redeployment
• Model testing automation

Learning Objectives:

• Implement CI/CD practices for ML workflows.
• Automate retraining, testing, and deployment of ML models.

Learning Outcomes:

• Able to use GitOps for ML deployment pipelines.
• Confident in automating model deployment lifecycle.

Tools Used:

• GitHub Actions
• Azure DevOps
• Docker

Subtopics:

• Deployment strategies: real-time, batch, A/B, canary
• Deploying to Azure Endpoints / Vertex AI Endpoints
• Model monitoring tools and metrics
• Detecting drift (data drift, concept drift)
• Re-training triggers and alerting

Learning Objectives:

• Deploy and monitor ML models in production environments.
• Implement robust model monitoring and drift detection.

Learning Outcomes:

• Able to deploy real-time and batch ML endpoints.
• Monitor and trigger retraining upon detecting drift.

Tools Used:

• Vertex AI Endpoints
• Azure Endpoints
• Prometheus
• Grafana

Subtopics:

• IAM & RBAC for ML workflows
• Securing endpoints and pipelines
• Secrets management (Key Vault / Secret Manager)
• Scaling with Kubernetes (intro only)
• Cost monitoring and optimization tips

Learning Objectives:

• Secure ML pipelines and manage access.
• Implement scalable and cost-effective ML infra.

Learning Outcomes:

• Able to manage IAM roles and secure secrets.
• Deploy models on scalable infra like Kubernetes.

Tools Used:

• Azure RBAC
• GCP IAM
• Key Vault
• Secret Manager

Subtopics:

• Introduction to Feature Stores
• Using Feast (GCP) or Azure Feature Store (if available)
• Data versioning with DVC / LakeFS
• Reproducibility challenges in ML

Learning Objectives:

• Use feature stores and manage versioned datasets.
• Enable reproducibility and consistency in data pipelines.

Learning Outcomes:

• Able to store and reuse features efficiently.
• Track and version data using specialized tools.

Tools Used:

• Feast
• DVC
• LakeFS

Subtopics:

• Choose from 2–3 real-world problem statements
• Full pipeline: Data → Training → CI/CD → Deployment → Monitoring
• Peer review and feedback

Learning Objectives:

• Apply MLOps concepts to solve real-world problems.
• Demonstrate end-to-end ML lifecycle management.

Learning Outcomes:

• Deliver a working ML solution using CI/CD and cloud deployment.
• Showcase project portfolio to potential employers.

Tools Used:

• All course tools

Subtopics:

• Resume building for MLOps roles
• Practice interview questions (technical + scenario)
• Mock interview session (optional)
• Certification exam guidance (Google Cloud ML, AI-102 Azure)

Learning Objectives:

• Prepare for MLOps job interviews and certification exams.
• Build a strong MLOps portfolio and resume.

Learning Outcomes:

• Confidently attend interviews and technical assessments.
• Tailor resume and project showcase for job applications.

Tools Used:

• LinkedIn
• Google Cert Prep
• Mock Interview Tools

Subtopics:

• Model Testing & Evaluation Frameworks (unit testing, integration tests)
• Bias Detection and Fairness Evaluation
• Introduction to Kubernetes and Kubeflow for ML Workloads
• Using EvidentlyAI and WhyLabs for Model Monitoring
• Advanced Data Validation with Great Expectations
• Scalability & Portability Considerations

Learning Objectives:

• Introduce advanced practices like bias detection, model validation, and infrastructure orchestration.
• Equip learners with tools used in production-grade MLOps environments.
• Learning Outcomes:
• Understand model validation strategies and bias checks.
• Gain exposure to Kubernetes, Kubeflow, and scalable ML infrastructure.
• Become familiar with lightweight tools for model drift detection and fairness.

Tools Used:

• Kubernetes
• Kubeflow
• EvidentlyAI
• WhyLabs
• Great Expectations
• Deepchecks