100+ Free CAIP (AIP-210) Practice Questions

Pass your CertNexus Certified Artificial Intelligence Practitioner (CAIP / AIP-210) exam on the first try — instant access, no signup required.

✓ No registration✓ No credit card✓ No hidden fees✓ Start practicing immediately

~60-70% Pass Rate

100+ Questions

100% Free

1 / 10

Question 1

Score: 0/0

Which type of machine learning uses labeled training data with known target outputs?

Unsupervised learning

Supervised learning

Reinforcement learning

Self-supervised learning

to track

2026 Statistics

Key Facts: CAIP (AIP-210) Exam

Exam Questions

CertNexus

120 min

Exam Duration

CertNexus

60-70%

Passing Score

CertNexus (scaled)

$400

Exam Fee

CertNexus

3 years

Validity

CEC renewal

Vendor-neutral

Format

All AI/ML stacks

The CAIP (AIP-210) exam has 80 questions in 120 minutes with a 60-70% scaled passing score. Five domains cover AI specifications and project lifecycle, data preparation and feature engineering, ML algorithm selection (regression, classification, clustering, ensembles), neural networks and deep learning (CNNs, RNNs, transformers), and operations/responsible AI (MLOps, fairness, NIST AI RMF, differential privacy). Exam fee is $400. Valid 3 years with CEC renewal.

Sample CAIP (AIP-210) Practice Questions

Try these sample questions to test your CAIP (AIP-210) exam readiness. Each question includes a detailed explanation. Start the interactive quiz above for the full 100+ question experience with AI tutoring.

1Which type of machine learning uses labeled training data with known target outputs?

A.Unsupervised learning

B.Supervised learning

C.Reinforcement learning

D.Self-supervised learning

Explanation: Supervised learning algorithms learn a mapping from inputs to known outputs (labels). Classification and regression are the two primary supervised tasks. Unsupervised learning finds structure in unlabeled data, and reinforcement learning learns from reward signals.

2What is the primary distinction between narrow AI and general AI (AGI)?

A.Narrow AI runs on smaller hardware

B.Narrow AI is task-specific; general AI matches human cognitive flexibility across tasks

C.Narrow AI is always rule-based

D.Narrow AI cannot use neural networks

Explanation: Narrow AI (ANI) is built for specific tasks (image classification, translation). Artificial General Intelligence (AGI) would match or exceed human cognitive flexibility across arbitrary domains and remains theoretical. All production AI today is narrow AI.

3Which task type predicts a continuous numeric value?

A.Classification

B.Regression

C.Clustering

D.Association rule mining

Explanation: Regression predicts continuous numeric values (price, temperature). Classification predicts discrete categories. Clustering groups similar items, and association rule mining finds frequent itemsets in transaction data.

4What does the train/validation/test split typically achieve?

A.It speeds up gradient descent

B.It separates data for fitting, hyperparameter tuning, and unbiased final evaluation

C.It removes outliers

D.It is required to compute accuracy

Explanation: Train data fits the model. Validation data tunes hyperparameters and selects the best model. Test data provides an unbiased estimate of generalization, and is touched only once. Common splits include 70/15/15 or 80/10/10.

5K-fold cross-validation is most useful when:

A.The training set is extremely large

B.Data is limited and you want a robust estimate of model performance

C.You want to deploy the model faster

D.Labels are noisy

Explanation: K-fold cross-validation rotates each fold as a validation set, averaging k performance estimates. It is most valuable when data is limited because every sample contributes to both training and validation across the folds.

6Which sampling technique preserves class proportions across train and test sets?

A.Random sampling

B.Stratified sampling

C.Cluster sampling

D.Systematic sampling

Explanation: Stratified sampling maintains the relative class frequencies in each split — critical when classes are imbalanced. Random sampling can produce splits where rare classes are absent from test or train data.

7SMOTE addresses class imbalance by:

A.Removing majority-class samples

B.Synthesizing new minority-class samples by interpolating between existing minority points and their nearest neighbors

C.Reweighting the loss function

D.Adding random noise to all features

Explanation: SMOTE (Synthetic Minority Over-sampling Technique) generates synthetic minority samples along line segments between a minority sample and its k nearest minority neighbors. It is preferred over random oversampling because it reduces overfitting on duplicates.

8Which imputation method is most appropriate for skewed numeric features?

A.Mean imputation

B.Median imputation

C.Mode imputation

D.Zero imputation

Explanation: Median is robust to skew and outliers, making it the better choice for non-normal numeric features. Mean is pulled by extreme values. Mode is for categorical features. Zero imputation can introduce systematic bias.

9Which advanced imputation method models each feature as a function of the others using iterative regression?

A.KNN imputation

B.MICE (Multiple Imputation by Chained Equations)

C.Mean imputation

D.Constant imputation

Explanation: MICE iteratively imputes each feature conditional on all others using regression. It captures multivariate dependencies and produces multiple plausible imputed datasets to reflect imputation uncertainty.

10Min-max normalization scales features to which range?

A.Mean 0, variance 1

B.0 to 1

C.-1 to 1

D.Sum to 1

Explanation: Min-max normalization scales x to (x - min) / (max - min), bounding values in [0, 1]. Standardization (z-score) gives mean 0 and variance 1. Both are useful when features have different scales.

About the CAIP (AIP-210) Exam

Certified Artificial Intelligence Practitioner (CAIP / AIP-210) is CertNexus's vendor-neutral applied AI certification. CAIP validates the ability to design, implement, and maintain AI/ML solutions across the lifecycle: problem framing, data preparation, feature engineering, algorithm selection, deep learning, evaluation, MLOps deployment, and responsible AI practices.

Questions

80 scored questions

Time Limit

120 minutes

Passing Score

60-70% (scaled)

Exam Fee

$400 USD (CertNexus / Pearson VUE)

CAIP (AIP-210) Exam Content Outline

~15%

AI Specifications and Project Lifecycle

Business problem framing, success criteria, data and constraint specifications, narrow vs general AI, supervised/unsupervised/reinforcement, and the end-to-end AI project lifecycle

~25%

Data Preparation and Feature Engineering

Data wrangling, EDA, train/test split, K-fold CV, stratified sampling, class imbalance (SMOTE/RUS/ROS), missing data imputation, normalization/standardization, encoding, PCA/t-SNE/UMAP, feature selection

~25%

ML Algorithm Selection

Linear/logistic regression, KNN, decision trees, ensembles (Random Forest, XGBoost, LightGBM, CatBoost), SVM, Naive Bayes, K-means, DBSCAN, hierarchical clustering, association rules

~20%

Neural Networks and Deep Learning

Perceptron, MLP, activations, optimizers (Adam, AdamW, RMSprop), regularization (dropout, BN, LN, early stopping), CNNs, transfer learning (ResNet, EfficientNet, ViT), RNN/LSTM/GRU, transformers

~15%

Operations and Responsible AI

Evaluation metrics (precision/recall/F1/ROC-AUC/MCC), MLOps (MLflow, Kubeflow, model registry), deployment patterns (REST/batch/streaming, A/B, canary, shadow), monitoring/drift, NIST AI RMF, differential privacy, federated learning, k-anonymity

How to Pass the CAIP (AIP-210) Exam

What You Need to Know

Passing score: 60-70% (scaled)
Exam length: 80 questions
Time limit: 120 minutes
Exam fee: $400 USD

Keys to Passing

Complete 500+ practice questions
Score 80%+ consistently before scheduling
Focus on highest-weighted sections
Use our AI tutor for tough concepts

CAIP (AIP-210) Study Tips from Top Performers

1Master the AI/ML lifecycle: business framing, data prep, modeling, evaluation, deployment, monitoring

2Know when to use each algorithm family: regression, trees, ensembles, SVM, neural networks

3Understand evaluation metrics deeply: precision, recall, F1, ROC-AUC, PR-AUC, MCC, and when each matters

4Practice handling class imbalance with SMOTE, class weights, and threshold tuning

5Learn MLOps fundamentals: MLflow tracking, model registry, deployment patterns, drift monitoring

6Study responsible AI: NIST AI RMF functions (Govern, Map, Measure, Manage), fairness, differential privacy

7Build an end-to-end project portfolio — CAIP rewards practical experience

Frequently Asked Questions

What is the CAIP (AIP-210) exam?

The Certified Artificial Intelligence Practitioner (CAIP / AIP-210) is CertNexus's vendor-neutral applied AI certification. It validates the practical skills required to design, implement, evaluate, and maintain AI and ML solutions across the project lifecycle. Topics include data preparation, feature engineering, classical ML, neural networks, MLOps, and responsible AI.

How many questions are on the CAIP exam?

The CAIP (AIP-210) exam has 80 questions to complete in 120 minutes. Questions are multiple-choice and scenario-based, focused on real-world model design and operational decisions. The passing score is scaled and typically corresponds to roughly 60-70% of items correct.

What is the largest domain on the CAIP exam?

Data Preparation and Feature Engineering and ML Algorithm Selection are typically the largest domains (each ~25%). Candidates should master EDA, handling missing data and class imbalance, PCA, scikit-learn pipelines, and ensemble methods like XGBoost and LightGBM.

Are there prerequisites for the CAIP exam?

There are no formal prerequisites. CertNexus recommends approximately 6+ months of hands-on Python ML experience, including scikit-learn and one deep learning framework (TensorFlow/Keras or PyTorch). Familiarity with statistics and linear algebra is helpful.

How long is CAIP valid?

CAIP certification is valid for 3 years from the date you pass. To renew, you must earn Continuing Education Credits (CECs) through training, conferences, publications, or professional activities and pay a CertNexus renewal fee.

How should I prepare for CAIP?

Plan for 60-100 hours of study over 6-10 weeks. Build hands-on projects covering EDA, feature engineering, classical ML (RF/XGBoost), and a deep learning project (CNN with transfer learning). Study NIST AI RMF and responsible AI concepts. Complete 100+ practice questions, scoring 80%+ before scheduling.