← Back to Dashboard
Module 07 · Days 43–49
MLOps & Production AI Systems

Getting a model to production is 90% of the work. Learn to deploy, monitor, scale, and maintain AI systems — the skills separating senior ML engineers from researchers.

D43
Model serving: FastAPI + Docker
Wrap any model as a production REST API. Input validation, health checks, latency logging.
D44
CI/CD for ML: GitHub Actions & DVC
Automated testing, linting, and deployment. Every merge that breaks a model gets caught before production.
D45
Kubernetes basics for ML engineers
Deployments, services, HPA, and GPU workloads. Enough to deploy and autoscale a model serving pod.
D46
Monitoring: drift detection, Prometheus & Grafana
Data drift, concept drift, Evidently for ML alerts, and dashboards that catch issues before users do.
D47
LLM serving: vLLM, TGI & inference optimisation
Serving LLMs in production. Continuous batching, PagedAttention, quantised inference — fast and cheap.
D48
Cloud ML platforms: AWS SageMaker & GCP Vertex AI
Managed training, endpoint deployment, autoscaling, and cost control on major cloud platforms.
🏗
🏗️ PROJECT: End-to-End MLOps Pipeline
Full system: model → FastAPI → Docker → GitHub Actions → Kubernetes → Evidently monitoring → Grafana dashboard.