MLOps Interview Prep
Test & Track Your Readiness

ML lifecycle, model types, evaluation metrics, and the ML engineering mindset

ETL patterns, feature engineering, feature stores, and data quality for ML

Distributed training, GPU clusters, hyperparameter tuning, and training pipelines

MLflow, experiment management, model versioning, and the model registry

Real-time vs batch inference, model serving patterns, containerization, and A/B testing

Data drift, concept drift, performance degradation, alerting, and retraining triggers

Kubeflow, SageMaker, Vertex AI, Azure ML — when to use which and how to evaluate

Model lineage, audit trails, explainability, fairness, and regulatory compliance

Why MLOps exists, Google's maturity model, three failure classes, ML lifecycle, and key personas

Feature stores, point-in-time correctness, Feast, Tecton, streaming features with Flink, and embedding stores

Distributed training with DDP and FSDP, Kubernetes for ML, pipeline orchestration, GPU optimization, and hyperparameter tuning

OCI containers for ML, ONNX cross-framework export, model registry promotion patterns, and model cards

FastAPI, Triton, TorchServe, vLLM, TGI, quantization, canary deployments, and serving SLOs

GitHub Actions pipelines, blue-green deployments, champion-challenger evaluation, rollback, and ML testing strategies

Great Expectations, Soda Core, PyDeequ, data contracts, schema evolution, and validation integration

PSI and KS drift detection, Evidently AI, prediction logging, SLOs, Prometheus/Grafana, and retraining triggers

DVC advanced patterns, Delta Lake time travel, Apache Iceberg, OpenLineage, and Marquez for lineage tracking

Terraform for ML infrastructure, Helm charts, Pulumi in Python, GitOps with ArgoCD, and cost management

Prompt versioning, RAG pipelines and evaluation, fine-tuning with LoRA/QLoRA, guardrails, and LLM observability

Three production ML systems: real-time fraud detection, content recommendation pipeline, and RAG Q&A with governance