Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Month 6 — Mashqlar to’plami

🟢 Easy

MLflow

  1. SQLite + MLflow + 5 ta run.
  2. mlflow.sklearn.autolog() use.
  3. Model Registry: register → Staging → Production.

DVC

  1. dvc init + bitta CSV versioning.
  2. Local DVC remote.
  3. 2 data versions, rolling back to the old version.

FastAPI Serving

  1. Sklearn modelni FastAPI to deploying.
  2. Health checks.
  3. Prometheus metrics.

Docker / K8s

  1. Multi-stage Dockerfile.
  2. Docker Compose: API + Postgres.
  3. minikube setup.

Monitoring

  1. Evidently AI birinchi report.
  2. PSI calculation.
  3. Custom Prometheus gauge.

CI/CD

  1. GitHub Actions pytest pipeline.
  2. Code quality checks.
  3. Docker build action.

Airflow

  1. Local Airflow Docker.
  2. First DAG (hello world).
  3. Daily scheduled task.

🟡 Medium

Integrations

  1. MLflow + DVC: ikkalasini birga loyihada.
  2. FastAPI + MLflow Registry: production from model yuklash.
  3. Docker Compose: API + MLflow + Postgres + MinIO.
  4. K8s + HPA: load test with auto-scaling.
  5. Airflow + MLflow: scheduled retraining DAG.

Real workflows

  1. Full retraining pipeline: DVC repro + MLflow log + K8s update.
  2. Daily inference batch: Airflow DAG, 100K users.
  3. Monitoring dashboard: Grafana + Prometheus + Evidently.
  4. A/B test: Istio or nginx canary deployment.
  5. CML report: Auto metrics comparison on PR.

🔴 Hard (Production)

1. End-to-End MLOps Platform

Requirements:

  • Classical ML model (regression or classification)
  • DVC for data versioning (S3 or MinIO)
  • MLflow for experiment tracking + Registry
  • DVC + MLflow integration
  • FastAPI serving + ONNX optimization
  • Docker + K8s deployment (manifest or Helm)
  • Prometheus + Grafana monitoring
  • Evidently AI drift detection
  • GitHub Actions CI/CD
  • Airflow scheduled retraining
  • Slack notifications

Deliverables:

  • GitHub repo (public)
  • README + architecture diagram
  • Demo video (Loom)
  • LinkedIn post

2. Multi-model Platform

Requirements:

  • 3+ different models (classification, regression, NLP)
  • Universal serving API (model-as-a-service)
  • Per-model routing and versioning
  • Centralized monitoring
  • Cost tracking per model/user
  • API rate limiting

3. Real-time Streaming ML

Requirements:

  • Kafka stream (or Redis Streams)
  • Real-time feature engineering
  • Low-latency inference (<50ms p95)
  • Online learning (River library)
  • Real-time monitoring dashboard

4. ML Platform as a Service (MLaaS)

Requirements:

  • User uploads CSV → auto-ML training
  • BentoML packaging
  • Auto-deployment to K8s
  • Per-user namespaces
  • Billing integration
  • Admin dashboard

Mini-projects

Mini-project 1: Personal Health ML Platform

  • Fitbit/Apple Health data
  • Predict health metrics
  • Daily inference + insights
  • Telegram bot

Mini-project 2: E-commerce Recommendation MLOps

  • Online learning (recommendations)
  • Feature store (Feast)
  • A/B test framework
  • Real-time deployment

Mini-project 3: Fraud Detection System

  • Streaming fraud detection
  • Real-time monitoring
  • Alert system
  • Explainability dashboard

Mini-project 4: Computer Vision SaaS

  • Multi-tenant CV API
  • Image moderation, OCR, classification
  • Usage tracking + billing
  • Streamlit demo

Quiz

MLOps Fundamentals

  1. Difference between MLOps and DevOps?
  2. The 8 stages of ML Lifecycle?
  3. MLOps Maturity Levels (0, 1, 2)?
  4. 3 main requirements for reproducibility?
  5. Why ML monitoring is harder than software monitoring?

MLflow

  1. Difference between Tracking, Models, Registry, Projects?
  2. How does auto-logging work?
  3. Model Registry stages workflow?
  4. How is a new model rolled out to production?
  5. MLflow vs W&B vs Neptune?

DVC

  1. Why is Git not enough for ML data?
  2. The role of dvc.yaml and dvc.lock?
  3. Remote storage options?
  4. Which stages does dvc repro re-run?
  5. DVC vs LakeFS vs Pachyderm?

Serving

  1. FastAPI custom vs BentoML vs TorchServe — when which?
  2. Why is batching important on GPU?
  3. Why is ONNX useful?
  4. Async inference patterns?
  5. Blue-green vs canary vs shadow deployment?

Docker / K8s

  1. Why multi-stage build?
  2. K8s Pod, Deployment, Service?
  3. Probes (liveness, readiness)?
  4. HPA by which metrics?
  5. What is KServe?

Monitoring

  1. Data drift, concept drift, prediction drift?
  2. PSI vs KS test?
  3. Evidently AI vs WhyLabs?
  4. Prometheus Counter vs Histogram vs Gauge?
  5. Retraining trigger logic?

CI/CD

  1. What’s added in ML CI/CD (vs classical DevOps)?
  2. Code, data, model testing?
  3. What does CML do?
  4. Deployment strategies?
  5. Rollback mechanism?

Airflow

  1. Difference between DAG and Task?
  2. Why XCom?
  3. Sensors?
  4. TaskFlow API vs traditional Operators?
  5. Airflow vs Prefect vs Dagster?

✅ Month 6 final checklist (the most important month!)

  • MLflow Tracking + Registry
  • DVC data versioning
  • FastAPI ML serving (production-ready)
  • Docker Compose stack
  • Local Kubernetes deployment
  • Prometheus + Grafana monitoring
  • Evidently AI drift detection
  • GitHub Actions ML pipeline
  • CML reports
  • Airflow DAG for retraining
  • End-to-end MLOps project on GitHub
  • Architecture diagram
  • LinkedIn post (certificate + GitHub link)
  • Update CV: “ML Engineer / MLOps Engineer”
  • Apply to 5+ vacancies

6 months complete!

You are now a complete ML Engineer / MLOps Engineer. Next step:

  1. Final Projects — 4 large projects for your portfolio
  2. Job applications — apply to vacancies
  3. Open source contributions — to MLflow, Evidently, DVC, etc.
  4. Speaking — talk about ML/MLOps at meetups
  5. Mentor — teach others

Everything is in your hands. Good luck!