Oy 6 — Mashqlar to'plami

🟢 Easy

MLflow

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

DVC

1. dvc init + bitta CSV versioning.
2. Local DVC remote.
3. 2 ta data versiya, eski versiyaga qaytish.

FastAPI Serving

1. Sklearn modelni FastAPI'ga olib chiqish.
2. Health checks.
3. Prometheus metric'lar.

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. Birinchi DAG (hello world).
3. Daily scheduled task.

🟡 Medium

Integrations

1. MLflow + DVC: ikkalasini birga loyihada.
2. FastAPI + MLflow Registry: production'dan model yuklash.
3. Docker Compose: API + MLflow + Postgres + MinIO.
4. K8s + HPA: load test bilan 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 yoki nginx canary deployment.
5. CML report: PR'ga avtomatik metrics comparison.

🔴 Hard (Production)

1. End-to-End MLOps Platform

Talab:

• Klassik ML modeli (regression yoki classification)
• DVC for data versioning (S3 yoki MinIO)
• MLflow for experiment tracking + Registry
• DVC + MLflow integration
• FastAPI serving + ONNX optimization
• Docker + K8s deployment (manifest yoki 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

Talab:

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

3. Real-time Streaming ML

Talab:

• Kafka stream (yoki 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)

Talab:

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

Mini-loyihalar

Mini-loyiha 1: Personal Health ML Platform

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

Mini-loyiha 2: E-commerce Recommendation MLOps

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

Mini-loyiha 3: Fraud Detection System

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

Mini-loyiha 4: Computer Vision SaaS

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

Quiz

MLOps Fundamentals

1. MLOps va DevOps farqi?
2. ML Lifecycle 8 bosqichi?
3. MLOps Maturity Levels (0, 1, 2)?
4. Reproducibility'ning 3 ta asosiy talabi?
5. Why ML monitoring is harder than software monitoring?

MLflow

1. Tracking, Models, Registry, Projects farqi?
2. Auto-logging qanday ishlaydi?
3. Model Registry stages workflow?
4. Production'ga yangi model qanday rollout qilinadi?
5. MLflow vs W&B vs Neptune?

DVC

1. Git nima uchun ML data uchun yetmaydi?
2. dvc.yaml va dvc.lock ning vazifasi?
3. Remote storage variantlari?
4. dvc repro qaysi stage'larni qayta ishga tushiradi?
5. DVC vs LakeFS vs Pachyderm?

Serving

1. FastAPI custom vs BentoML vs TorchServe — qaysi qachon?
2. Batching nima uchun GPU'da muhim?
3. ONNX nima uchun foydali?
4. Async inference patternlari?
5. Blue-green vs canary vs shadow deployment?

Docker / K8s

1. Multi-stage build nima uchun?
2. K8s Pod, Deployment, Service?
3. Probes (liveness, readiness)?
4. HPA qaysi metric'lar bo'yicha?
5. KServe nima?

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. ML CI/CD da nima qo'shimcha (klassik DevOps'ga nisbatan)?
2. Code, data, model testing?
3. CML nima qiladi?
4. Deployment strategies?
5. Rollback mechanism?

Airflow

1. DAG va Task farqi?
2. XCom nima uchun?
3. Sensor'lar?
4. TaskFlow API vs traditional Operators?
5. Airflow vs Prefect vs Dagster?

✅ Oy 6 oxiri checklist (eng muhim oy!)

• 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 loyiha GitHub'da
• Architecture diagram
• LinkedIn post (sertifikat + GitHub link)
• CV'ni yangilash: "ML Engineer / MLOps Engineer"
• 5+ vakansiyaga ariza yuborish

6 oy tugadi!

Siz endi to'liq ML Engineer / MLOps Engineersiz. Keyingi bosqich:

1. Final Loyihalar — portfolio uchun 4 katta loyiha
2. Job applications — vakansiyalarga ariza
3. Open source contributions — MLflow, Evidently, DVC, va h.k. ga
4. Speaking — meetup'larda ML/MLOps haqida gapirish
5. Mentor — boshqalarga o'rgatish

Hamma narsa sizning qo'lingizda. Omad!