AI/ML Screening TestBy Vikas Sharma / 19/08/2025 AI/ML Screening Test 1 / 49 What is blue-green deployment in ML pipelines? Splitting training datasets randomly Maintaining two identical environments (blue and green) to switch traffic safely during updates Running models in GPUs only Using two ML algorithms simultaneously 2 / 49 Which monitoring metric is MOST relevant in MLOps? Number of Git commits Model accuracy and drift detection Website traffic CPU utilization only 3 / 49 Which is a key output of anomaly detection in AIOps? Identified unusual events that may indicate system issues CI/CD deployment reports Application code coverage Optimized hyperparameters 4 / 49 Which of the following best describes model governance Processes ensuring compliance, auditability, and security in ML models Hyperparameter optimization Anomaly detection only Visualization dashboards 5 / 49 What is a key advantage of using AIOps in incident management? Replacing monitoring tools entirely Proactive anomaly detection and root cause analysis Manual intervention for faster resolutions Increased number of false alerts 6 / 49 What is the role of GitOps in MLOp? Managing ML infrastructure and deployments declaratively through Git Running hyperparameter optimization Training ML models Visualizing anomalies 7 / 49 Which cloud service provides a fully managed ML pipeline solution? VMware vSphere AWS SageMaker Pipelines Photoshop Cloud Kubernetes without ML 8 / 49 . What is shadow deployment in MLOps? Deploying only half the model Running a new model in parallel with the current one without serving predictions to users Deploying without monitoring Deploying on shadow servers only 9 / 49 Why is explainability important in production ML models? To reduce deployment frequency To reduce CI/CD runtime To understand model decisions and build trust with stakeholders To increase data size 10 / 49 In a CI/CD pipeline, unit tests for ML models typically validate: Data preprocessing and feature transformations Operating system drivers Network bandwidth User interface design 11 / 49 Which of the following tools integrates monitoring into MLOps pipelines? Prometheus & Grafana PowerPoint Slack Tableau only 12 / 49 What is the main purpose of MLOps? To replace software engineering practices To automate cloud billing processes To build web applications To integrate ML models into production through CI/CD pipelines 13 / 49 What is the role of continuous validation in MLOps Improves GPU performance Reduces network traffic Ensures deployed models remain accurate and reliable with new data Tracks Git commits 14 / 49 What is Canary Deployment in MLOps? Deploying models only in staging Gradually rolling out a model to a subset of users before full release Deploying multiple models in parallel permanently Deploying models without validation 15 / 49 Which of the following ensures reproducibility in ML experiments? Versioning code, data, and models Manual hyperparameter tuning only Skipping documentation Avoiding CI/CD 16 / 49 . What role does Natural Language Processing (NLP) play in AIOps? Creating CI/CD pipelines Training computer vision models Provisioning infrastructure Parsing log files and correlating incidents 17 / 49 Why is monitoring critical after model deployment? To reduce developer workload To speed up CI builds only To reduce hardware costs To detect performance degradation and drift 18 / 49 What is the difference between DevOps and MLOps? DevOps focuses on CI/CD for software, while MLOps extends it to ML models with added steps like training and monitoring MLOps is only about data collection DevOps is only for cloud computing MLOps replaces DevOps entirely 19 / 49 What is the main role of Docker in MLOps pipelines? To containerize ML models for consistent deployment To perform hyperparameter tuning To analyze log anomalies To act as a monitoring dashboard 20 / 49 How does AIOps reduce 'alert fatigue? By generating more alerts By automating deployments only By correlating events and suppressing noise By disabling monitoring tools 21 / 49 What does CI/CD integration with model registry achieve? Tracks GitHub issues only Simplifies HTML rendering Improves IDE performance Automates promotion of validated models to production 22 / 49 What is the purpose of a model registry in MLOps? To manage Kubernetes clusters To store cloud infrastructure templates To track CI/CD pipeline executions To store, version, and manage trained ML models 23 / 49 Which of the following ensures fairness and bias detection in ML models? Relying on accuracy only Responsible AI practices and monitoring Skipping validation Using random data 24 / 49 In MLOps, what is 'model lineage? Measuring network latency Monitoring server uptime Tracking datasets, code, and parameters that produced a model Versioning HTML files 25 / 49 What is the purpose of data drift detection? To identify changes in input data distribution affecting model performance To detect server failures To version-control datasets To optimize CI/CD runtime 26 / 49 Which of the following is NOT a stage in the MLOps lifecycle? Model destruction Model training Model deployment Model monitoring 27 / 49 What is the purpose of MLflow in MLOps? Log analysis Container orchestration Experiment tracking, model registry, and deployment Database sharding 28 / 49 Which tool is widely used for managing ML pipelines? Nagios Jenkins Terraform Kubeflow 29 / 49 Which of the following best describes the goal of AIOps? Replacing DevOps entirely Applying AI/ML techniques to IT operations for proactive issue detection Automating CI/CD pipelines without monitoring Automating infrastructure scaling only 30 / 49 Which challenge does AIOps primarily address? Limited access to GitHub repositories Inability to run unit tests Manual analysis of large-scale operational data Lack of cloud cost optimization 31 / 49 Which type of data is MOST commonly analyzed by AIOps platforms? Structured business data Customer satisfaction surveys Unstructured IT operations data like logs, metrics, and traces Video and image datasets 32 / 49 What is online learning in ML deployment Batch scoring only Deploying only during office hours Offline retraining every month Updating the model incrementally with streaming data 33 / 49 What is the role of Kubernetes in MLOps pipelines Data preprocessing Model evaluation Scaling and orchestrating ML workloads in production Hyperparameter tuning only 34 / 49 Which of the following tools is commonly associated with AIOps? Moogsoft Apache Spark Terraform Kubernetes 35 / 49 Which of the following is an example of predictive analytics in AIOps? Forecasting disk failures before they occur Manual root cause analysis Static capacity planning Real-time log streaming 36 / 49 Which algorithm is often used in AIOps for log anomaly detection? Static Regex Matching Naive Bayes only Decision Trees for UI LSTM (Long Short-Term Memory) networks 37 / 49 In MLOps, what is 'model drift'? When model performance degrades due to changes in data patterns When the model is moved between servers When models crash during deployment When hyperparameters remain constant 38 / 49 . Which tool is commonly used for workflow orchestration in ML pipelines? Jenkins only Apache Airflow Nagios Excel 39 / 49 Which orchestrator is commonly used for ML pipelines in Kubernetes? Splunk Nagios Kubeflow Pipelines Airflow only 40 / 49 Which metric is best for evaluating classification models in imbalanced dataset? CPU usage Mean Squared Error Precision-Recall AUC Accuracy only 41 / 49 Which CI/CD tool is widely integrated with MLOps pipelines? Photoshop Jenkins MS Word Final Cut Pro 42 / 49 What is a common challenge in automating ML pipelines? Automating UI testing Writing HTML code Cloud billing alerts Data versioning and reproducibility 43 / 49 What is 'model rollback' in CI/CD pipelines Resetting hyperparameters Restarting the server Re-training from scratch Reverting to a previous stable model when the new one fails 44 / 49 Which of the following is an example of CI/CD for ML models? Running experiments locally only Manual model validation Automating retraining, testing, and deployment of models Skipping version control 45 / 49 Which of the following is a common model deployment pattern? Round-Robin Compilation Static Scaling Git Rebase Deployment Blue-Green Deployment 46 / 49 . What does a feature store provide in MLOps? A monitoring dashboard A CI/CD orchestrator A centralized repository for storing and sharing ML features A code versioning platform 47 / 49 Which AI technique is commonly used in AIOps for anomaly detection? Manual log parsing Clustering algorithms Linear regression only Rule-based filtering 48 / 49 Which stage in MLOps involves hyperparameter tuning? Incident management Deployment Model training & optimization Monitoring 49 / 49 Which of the following describes Continuous Training (CT) in MLOps? Deploying models continuously without validation Scaling infrastructure on demand Re-training models regularly with new data Running unit tests for ML code Your score is Share this: Share on Facebook (Opens in new window) Facebook Share on X (Opens in new window) X