Role-specific and topic-specific questions with answers
Easy, medium, and hard questions that cover most topics in machine learning and data science interviews. Solutions that deep dive into explaining complicated concepts, with necessary references and simulations when needed.
| Problem | Role | Area | Topics | Difficulty | Company (login required) | Status |
|---|---|---|---|---|---|---|
| Sample Ratio Mismatch | DS PDS AS | A/B Testing | Sample Ratio Mismatch | Medium | Duolingo Snap Uber | |
| Questions worksheet (recruiter-screen) | DS AS MLE PDS | Application process | Discussion prep, Application prep | Medium | Meta Robinhood Tik Tok | |
| Questions worksheet (behavioral) | DS AS MLE PDS | Behavioral | Discussion prep | Hard | Chan-Zuckemberk Discord Uber | |
| Minimum remove to make valid parentheses (Leetcode 1249) | DS AS MLE | Data Structures and Algorithms | Stack | Medium | Amazon | |
| Remove duplicates in place (Leetcode 26) | DS AS MLE | Data Structures and Algorithms | Array, Two pointers | Easy | ||
| Bias-variance equation | DS AS MLE | Machine Learning | Bias-variance tradeoff, Formula derivation | Easy | Fidelity Google Salesforce Workiva | |
| Linear regression with gradient descent | DS AS MLE | Machine Learning Coding | Gradient descent, Linear regression | Medium | Apple Hinge Uber | |
| Gradient descent vs. stochastic gradient descent | DS AS MLE | Machine Learning | Gradient descent, Stochastic gradient descent, Minibatch | Medium | Paypal | |
| Imbalanced dataset | PDS DS AS MLE | Machine Learning | Class imbalance | Medium | Discord Paypal Salesforce | |
| Sudden drop in user engagement | DS PDS | Metrics | Problem solving, Root-cause analysis | Medium | Snap | |
| Documents edited by AI | DS PDS AS | Probability theory | Bayes rule, Conditional probability | Easy | Indeed Snap Thumbtack Warner Bros | |
| Trailing by two: should we go for two or three? | DS PDS AS | Probability theory | Independence, Decision making | Easy | Microsoft | |
| Artist maxranks | DS AS PDS | SQL | Join | Easy | Paypal | |
| Two-sample t-test | DS PDS AS | Statistics | Hypothesis testing | Easy | ||
| Choose a project | DS AS MLE PDS | Technical deep dive | Discussion prep | Hard | ||
| Normality assumption | DS PDS AS | A/B Testing | Normality | Medium | ||
| Reducing variance in AB testing | DS PDS AS | A/B Testing | Variance | Medium | Apple | |
| Simpson’s paradox | DS PDS AS | A/B Testing | Simpson’s paradox | Medium | Fidelity Snap | |
| Interview success probability | DS AS MLE PDS | Application process | Application prep | Easy | ||
| Offer validity and cooling-off periods | DS AS MLE PDS | Application process | Application prep | Easy | ||
| Referrals vs. online applications | DS AS MLE PDS | Application process | Application prep | Easy | ||
| Response time | DS AS MLE PDS | Application process | Application prep | Easy | ||
| Resume review | DS AS MLE PDS | Application process | Application prep, Resume | Easy | ||
| Timeline: from application to offer | DS AS MLE PDS | Application process | Application prep | Easy | ||
| Up-level during interview | DS AS MLE PDS | Application process | Application prep | Easy | ||
| K Closest Points to Origin (Leetcode 973) | DS AS MLE | Data Structures and Algorithms | Heap | Easy | ||
| Longest substring without repeating characters (Leetcode 3) | DS AS MLE | Data Structures and Algorithms | Hash, Sliding Window | Medium | Amazon | |
| Number of recent calls (Leetcode 933) | DS AS MLE | Data Structures and Algorithms | Queue, DEQueue | Easy | ||
| Analyze prediction error | DS AS MLE | Machine Learning | Prediction error, Bias-variance tradeoff, Diagnostics, Learning curves | Medium | Google Salesforce Snap Thumbtack | |
| AUC ROC and predicted output transformations | DS AS MLE | Machine Learning | AUC ROC | Easy | ||
| Sample from random generator | DS AS MLE | Machine Learning Coding | Sample, Uniform, Random number generator | Medium | Google Snap | |
| Simulate dynamic coin flips | DS AS MLE | Machine Learning Coding | Simulation | Easy | Cruise | |
| Encoding categorical features | PDS DS AS MLE | Machine Learning | Categorical features, Embeddings, One-hot encoding, Hashing | Easy | Apple Hinge Thumbtack | |
| Example of high-bias and high variance | DS AS MLE | Machine Learning | Bias-variance tradeoff | Medium | ||
| k-means | PDS DS AS MLE | Machine Learning | k-means | Medium | ||
| L1 (Lasso) vs. L2 (Ridge) regularization | DS AS MLE | Machine Learning | L1, L2, Redularization, Lasso, Ridge | Medium | Instacart Uber | |
| Linear regression likelihood function | DS AS MLE | Machine Learning | Linear regression, Likelihood function, Formula derivation | Medium | Uber | |
| Model interpretability | DS AS MLE | Machine Learning | Interpretability | Easy | Paylocity | |
| Multiclass evaluation metrics | DS AS MLE | Machine Learning | Multiclass, Diagnostics | Medium | Fidelity Paypal | |
| NDCG vs. Mean Average Precision (MAP) vs. Mean Reciprocal Rank (MRR) | AS MLE | Machine Learning | Ranking metrics, NDCG, Mean average precision (MAP), Mean Reciprocal Rank (MRR), Recommender Systems | Medium | Apple Coursera Indeed Paypal | |
| Range of R2 when combining regressions | PDS DS AS MLE | Machine Learning | Linear regression, Goodness of fit, R-squared, Correlation | Medium | D.E. Shaw | |
| ROC vs. PR curve | AS MLE | Machine Learning | AUC, ROC, AUPR, Precision, Recall, Evaluation metrics | Medium | Amazon Google Meta Reddit | |
| Measuring counterfactual impact | DS PDS | Metrics | Problem solving, Counterfactual | Hard | Meta Snap | |
| Games between two players | DS PDS AS | Probability theory | Recursive relationship | Medium | Amazon | |
| Monty Hall | DS PDS AS | Probability theory | Bayes rule, Conditional independence, Prior evidence | Medium | ||
| Red and blue balls | DS PDS AS | Probability theory | Counting, Combinations, Repetition, Binomial | Easy | Indeed | |
| Sum of normally distributed random variables | DS PDS AS | Probability theory | Normal, PDF, CDF | Medium | Meta | |
| Two children I | DS PDS AS | Probability theory | Prior evidence | Easy | Amazon | |
| Unfair coin probability | DS PDS AS | Probability theory | Bayes rule, Conditional probability | Easy | ||
| Choose house or techno | DS AS PDS | SQL | Logical OR | Easy | ||
| Expensive house songs I | DS AS PDS | SQL | Subquery, CTE, Join | Medium | ||
| Expensive house songs II | DS AS PDS | SQL | Subquery, CTE, Join, Window functions | Hard | Google Snap Tik Tok | |
| Histogram of songs | DS AS PDS | SQL | Recursive CTE, Join | Hard | ||
| Songs that did not enter the charts or entered high | DS AS PDS | SQL | Subquery, Join | Medium | ||
| Songs that ranked 1 to 50 | DS AS PDS | SQL | Between | Easy | ||
| Songs that stay in the chars for a while | DS AS PDS | SQL | Subquery, CTE, Join, ALL, Window functions | Medium | ||
| Biased coin | DS PDS AS | Statistics | Expectation, CLT, Binomial, Normal, Bernoulli, Hypothesis testing, CDF | Medium | Meta | |
| Gambler’s ruin win probability | DS PDS AS | Statistics | Gambler ruin, Random walk, Expectation | Medium | Meta | |
| Manual estimation of flips | DS PDS AS | Statistics | Normal, CDF, Binomial, CLT | Medium | ||
| Measuring sticks | DS PDS AS | Statistics | Variance | Medium | Sisu | |
| Monotonic draws | DS PDS AS | Statistics | Expectation | Hard | ||
| Prussian horses | DS PDS AS | Statistics | Poisson, Hypothesis testing, CDF | Medium | Indeed | |
| Relationship between p-val and confidence interval | DS PDS AS | Statistics | Confidence interval, P-value, Hypothesis testing | Easy | ||
| Questions worksheet (deep dive) | DS AS MLE PDS | Technical deep dive | Discussion prep | Hard | ||
| AA tests | DS PDS AS | A/B Testing | Variance | Easy | Apple Snap Uber | |
| Counterfactual definition | DS PDS AS | A/B Testing | Counterfactual | Easy | ||
| Equal-sized treatment and control groups | DS PDS AS | A/B Testing | Power, Variance, Sample size | Medium | Snap | |
| False discovery control | DS PDS AS | A/B Testing | False discovery rate, Multiple hypotheses testing, Benjamini & Hochberg, Bonferroni | Easy | Discord Pinterest | |
| Interference | DS PDS AS | A/B Testing | Interference | Easy | Snap TaskRabbit Uber | |
| Multi-armed and contextual bandits in AB testing | DS PDS AS | A/B Testing | Contextual bandits, Multi-armed bandits | Medium | ||
| Novelty and primacy effects | DS PDS AS | A/B Testing | Novelty effects, Primacy effects | Easy | Uber | |
| Randomization checks | DS PDS AS | A/B Testing | Randomization | Easy | ||
| Randomization level | DS PDS AS | A/B Testing | Randomization, Variance | Medium | ||
| Climbing stairs (Leetcode 70) | DS AS MLE | Data Structures and Algorithms | Recursion, Dynamic programming | Easy | ||
| Find if Path Exists in Graph (Leetcode 1971) | DS AS MLE | Data Structures and Algorithms | DFS, BFS | Medium | Amazon | |
| Search in Binary Search Tree (Leetcode 700) | DS AS MLE | Data Structures and Algorithms | Binary Search, Binary Search Tree | Easy | ||
| Sort an Array (Leetcode 912) | DS AS MLE | Data Structures and Algorithms | Recursion, Sorting | Medium | ||
| Activation functions | AS MLE | Machine Learning | Neural networks, Deep learning | Easy | Cruise | |
| Active learning | AS MLE | Machine Learning | Labels, Label sampling | Medium | Dropbox | |
| Adagrad vs. RMSProp vs. Adam | AS MLE | Machine Learning | Neural networks, Deep learning, Optimization | Hard | Apple Cruise Instacart | |
| Adaptive learning rate | AS MLE | Machine Learning | Neural networks, Deep learning, Optimization | Hard | Cruise Instacart Two Sigma | |
| Approximate nearest neighbors | AS MLE | Machine Learning | ANN, ANNOY, Nearest neighbors | Medium | Dropbox Lacework | |
| Attention (intuition) | AS MLE | Machine Learning | Neural networks, Deep learning, Transformers, LLMs | Easy | Hinge Lacework Tik Tok | |
| Baselines | AS MLE DS | Machine Learning | Model evaluation | Easy | ||
| Bayesian frequentist statistics | AS MLE | Machine Learning | Bayesian, Frequentist | Easy | ||
| Bias-variance biased estimator | DS AS MLE | Machine Learning | Bias-variance tradeoff | Medium | ||
| Bootstrap | PDS DS AS MLE | Machine Learning | Bootstrap | Easy | Google LinkedIn | |
| K-means from scratch | DS AS MLE | Machine Learning Coding | k-means | Medium | Amazon Etsy Snap | |
| Linear regression with stochastic gradient descent | DS AS MLE | Machine Learning Coding | Stochastic Gradient descent, Linear regression | Medium | ||
| Logistic regression with gradient descent | DS AS MLE | Machine Learning Coding | Gradient descent, Logistic regression | Medium | ||
| Naive Bayes from scratch | DS AS MLE | Machine Learning Coding | Gaussian Naive Bayes | Medium | Hinge | |
| Neural network implementation | AS MLE | Machine Learning Coding | Gradient descent, Neural networks, Neuron | Hard | Uber | |
| Principal Component Analysis (PCA) from scratch | DS AS MLE | Machine Learning Coding | Principal Component Analysis (PCA) | Medium | ||
| Common causes of data leakage | DS AS MLE | Machine Learning | Data leakage | Medium | ||
| Comparing decision trees with random forests | DS AS MLE | Machine Learning | Decision trees, Random forests | Easy | Paypal | |
| Correlation with binary variables | DS AS MLE | Machine Learning | Correlation, Hypothesis testing, Point-biserial correlation coefficient | Easy | Meta | |
| Cross validation | PDS DS AS MLE | Machine Learning | Cross validation, Offline evaluation | Easy | Amazon Amperity Discord | |
| Decide between a multinomial vs. a binary modeling approach | AS MLE | Machine Learning | Modeling, Multinomial, Binary | Easy | ||
| Discretization drawbacks | DS AS MLE | Machine Learning | Categorical variables, Discretization | Easy | ||
| Ensembles | AS MLE DS | Machine Learning | Ensembles, Numerical example | Medium | ||
| Examples of encoder and decoder models | AS MLE | Machine Learning | LLMs, Transformers, Encoder, Decoder | Easy | Fidelity Salesforce | |
| Exponentially weighted moving average | AS MLE | Machine Learning | Exponentially weighted moving average, Formula derivation, Proof | Medium | ||
| Feature crossing | AS MLE | Machine Learning | Feature engineering, Deep learning | Easy | Meta | |
| Feature engineering in the era of deep learning | DS AS MLE | Machine Learning | Feature engineering | Easy | ||
| Gini impurity vs. information gain | AS MLE | Machine Learning | Decision tree, Information Gain, Gini impurity | Medium | ||
| Gradient boosting vs. random forests | DS AS MLE | Machine Learning | Gradient boosting, Random forests, Bagging, Boosting | Medium | Indeed Instacart Meta Salesforce | |
| Gradient descent vs. Stochastic Gradient descent and local minima | AS MLE | Machine Learning | Gradient descent (GD), Stochastic gradient descent (SGD), Local minima, Optimization, Deep learning | Medium | Cruise Discord Google Indeed | |
| Gradient descent vs. Stochastic Gradient descent and learning rate | AS MLE | Machine Learning | Gradient descent (GD), Stochastic gradient descent (SGD), Learning rate, Optimization, Deep learning | Medium | Paypal | |
| Predict whether a movie will receive good reviews | AS MLE | Machine Learning Hands on | Feature engineering, Data exploration, ML modeling, Logistic regression, One hot encoding | Hard | ||
| How to get more labels | AS MLE | Machine Learning | Modeling, Label encoding | Medium | ||
| Hypothesis testing in regression coefficients | DS AS MLE | Machine Learning | Linear regression, Hypothesis testing | Medium | ||
| Information gain in decision trees | AS MLE | Machine Learning | Decision tree, Entroy, Information Gain, Formula derivation | Medium | ||
| Intercept | PDS DS AS MLE | Machine Learning | Linear regression, Intercept | Easy | ||
| Interpretability | PDS DS AS MLE | Machine Learning | ML interpretability | Easy | Apple | |
| L2 regularization vs. weight decay | AS MLE | Machine Learning | Neural networks, Deep learning, Regularization | Hard | Apple Google Instacart ROKT | |
| Linear regression assumptions | PDS DS AS MLE | Machine Learning | Linear regression | Easy | Fidelity | |
| Linear regression with duplicated rows | DS AS MLE | Machine Learning | Linear regression, Statistical significance | Easy | ||
| Linear regression with stochastic gradient descent (formula derivation) | AS MLE | Machine Learning | Linear regression, Stochastic gradient descent, Formula derivation | Medium | ||
| Logistic regression and standardization | DS AS MLE | Machine Learning | Logistic regression, Standardization | Easy | Indeed Paypal | |
| Logistic regression assumptions | PDS DS AS MLE | Machine Learning | Logistic regression | Easy | Indeed Paypal Warner Bros | |
| Minimization of loss function intuition | AS MLE | Machine Learning | Neural networks, Deep learning, Optimization | Easy | Robinhood Uber | |
| Missing data | PDS DS AS MLE | Machine Learning | Missing data | Easy | Apple Flatiron Health | |
| Momentum | AS MLE | Machine Learning | Neural networks, Deep learning, Optimization | Hard | Apple Cruise Instacart Paypal | |
| MSE vs. MAE | PDS DS AS MLE | Machine Learning | MSE, MAE | Easy | ||
| Multi-headed attention and self attention | AS MLE | Machine Learning | Neural networks, Deep learning, Transformers, LLMs | Medium | Amazon Google | |
| Multicollinearity | PDS DS AS MLE | Machine Learning | Multicollinearity, Linear regression | Medium | Paypal | |
| Negative sampling | AS MLE | Machine Learning | Neural networks, Deep learning, Negative sampling | Medium | Dropbox Reddit | |
| Neuaral networks in layman terms | AS MLE | Machine Learning | Neural networks, Deep learning | Easy | Fidelity | |
| Non-probability sampling | DS AS MLE | Machine Learning | Sampling, Non-probability | Easy | ||
| Normalization in neural networks | AS MLE | Machine Learning | Neural networks, Deep learning, Batch normalization, Layer normalization | Medium | Amazon Google Tinder | |
| Normalization vs. Standardization | PDS DS AS MLE | Machine Learning | Linear regression, Standardization, Normalization | Easy | ||
| Not enough data to train a model | DS AS MLE | Machine Learning | Data limitations | Easy | ||
| Optimize multiple objectives | AS MLE | Machine Learning | Modeling, Multiple objectives | Easy | ||
| Outliers | PDS DS AS MLE | Machine Learning | Outliers, Cook’s distance, Regularization | Easy | ||
| Overfitting in neural networks | AS MLE | Machine Learning | Neural networks, Deep learning, Overfitting | Medium | Apple Google Indeed Instacart | |
| Positional embeddings | AS MLE | Machine Learning | Feature engineering, Deep learning, Transformers, LLMs, Positional embeddings, Positional encodings | Hard | ||
| Principal Component Analysis (PCA) | PDS DS AS MLE | Machine Learning | PCA | Easy | ||
| Prove that a median minizes MAE | AS MLE | Machine Learning | MAE, Median, Formula derivation, Proof | Hard | LinkedIn Uber | |
| Random forest feature importance | AS MLE | Machine Learning | Feature importance, Explainability, Gini importance, Permutation importance | Medium | Discord Grammarly Hinge | |
| Random vs. stratified sampling | PDS DS AS MLE | Machine Learning | Sampling, Stratified sampling | Easy | Meta | |
| Self-supervised learning | AS MLE | Machine Learning | Neural networks, Deep learning, Contrastive learning | Easy | Dropbox Reddit | |
| SMOTE | AS MLE | Machine Learning | Imbalanced classification, SMOTE, Data augmentation | Easy | Paypal Robinhood Snap Thumbtack | |
| API patterns | MLE | Machine Learning System Design | APIs, GraphQL, REST | Easy | Lacework | |
| Build an ML system to predict Ad clicks | AS MLE | Machine Learning System Design | ML system design, Feature engineering, Data exploration, ML modeling, Monitoring, Deployment, Business metrics | Hard | Meta | |
| Cloud vs. on-device deployment | MLE | Machine Learning System Design | Deployment, Cloud, Edge | Medium | ||
| Complex vs. simple deployment | MLE | Machine Learning System Design | Deployment | Easy | ||
| Crons, schedulers, orchestrattors | MLE | Machine Learning System Design | ML infra | Medium | Dropbox | |
| Data, model, and pipeline parallelism | MLE | Machine Learning System Design | Parallelism | Medium | ||
| Debug an ML model | MLE | Machine Learning System Design | Best practices | Medium | ||
| How to speed up inference | MLE | Machine Learning System Design | Inference | Easy | Dropbox Grammarly Robinhood Uber | |
| ML system design tools and use cases | MLE | Machine Learning System Design | ML infra, CDN, Kafka, Reddis, Dynamo, Cassandra, Chubby, PGVector, DBT, Feast, MLFlow, Statsig, Airflow, Fiddler | Hard | Hinge Reddit | |
| Online prediction, vs. batch prediction | MLE | Machine Learning System Design | Inference | Medium | Dropbox Reddit Uber | |
| Simple model deployment process | MLE | Machine Learning System Design | Deployment, Docker | Easy | ||
| Training tracking | MLE | Machine Learning System Design | Best practices | Medium | ||
| Types of data distribution shifts | MLE | Machine Learning System Design | Train-serving skew, Covariate shift, Label shift, Concept shift | Medium | Dropbox Lacework | |
| Transfer learning | AS MLE | Machine Learning | Neural networks, Deep learning, Transformers, LLMs, Catastrophic forgetting | Easy | ||
| Transfer learning vs. knowledge distillation | AS MLE | Machine Learning | LLM, Deep learning, Transfer learning, Knowledge distillation | Medium | ||
| Vanishing and exploding gradients (mathematical explaination) | AS MLE | Machine Learning | Neural networks, Deep learning, Mathematical explaination | Hard | Robinhood Salesforce | |
| Weight initialization | AS MLE | Machine Learning | Neural networks, Deep learning | Medium | Tinder | |
| Weighted and importance sampling | DS AS MLE | Machine Learning | Sampling, Weighted sampling, Importance sampling | Easy | ||
| Characteristics of metrics | DS PDS | Metrics | Characteristics of metrics | Easy | Duolingo Snap | |
| Types of metrics | DS PDS | Metrics | Types of metrics | Easy | ||
| Consecutive tails | DS PDS AS | Probability theory | Permutations, Repetition | Easy | ||
| Largest number rolled | DS PDS AS | Probability theory | Counting, Permutations, Repetition | Medium | ||
| Median probability | DS PDS AS | Probability theory | Binomial, Uniform, CDF | Medium | Meta | |
| Number of emails | DS PDS AS | Probability theory | Poisson distribution | Easy | Cruise | |
| Paths to destination | DS PDS AS | Probability theory | Counting, Combinations | Easy | ||
| Repeated rolls until 4 | DS PDS AS | Probability theory | Geometric distribution | Easy | ||
| Sample digits 1-10 | DS PDS AS | Probability theory | Sample from samples, Uniform | Medium | Compass Google Snap | |
| Two fair die rolls | DS PDS AS | Probability theory | Independence, CDF, PMF | Easy | ||
| Artists with more songs than others | DS AS PDS | SQL | Subquery, CTE, Join, Window functions | Hard | Meta Snap Tik Tok | |
| Concat columns | DS AS PDS | SQL | Concat | Easy | ||
| Label recent songs | DS AS PDS | SQL | Case | Easy | ||
| Median songs per artist | DS AS PDS | SQL | CTE, Window functions | Hard | ||
| Songs in charts with greater durations | DS AS PDS | SQL | Subquery, CTE, Join, Window functions | Hard | ||
| Songs per genre | DS AS PDS | SQL | Group by | Easy | ||
| Songs with letters | DS AS PDS | SQL | Regexp | Easy | ||
| An intuitive way to write power | DS PDS AS | Statistics | Power, Hypothesis testing | Easy | ||
| Buy and sell stocks | DS PDS AS | Statistics | Gambler ruin, Expectation, Recursion, Random walk | Medium | D.E. Shaw | |
| CI of flipping heads | DS PDS AS | Statistics | Confidence Interval, CLT, Bernoulli trials | Medium | ||
| Confidence interval definition | DS PDS AS | Statistics | Confidence interval, Hypothesis testing | Easy | Meta | |
| Confidence intervals that overlap | DS PDS AS | Statistics | Confidence interval, Hypothesis testing | Medium | ||
| Covariance of dependent variables | DS PDS AS | Statistics | Variance, Uniform, Covariance, Expectation | Medium | ||
| Distribution of a CDF | DS PDS AS | Statistics | CDF, Inverse transform | Medium | ||
| Dynamic coin flips | DS PDS AS | Statistics | Expectation, Simulation | Hard | Cruise | |
| Expected number of consecutive heads | DS PDS AS | Statistics | Expectation | Medium | ||
| Number of draws to get greater than 1 | DS PDS AS | Statistics | Normal, Geometric, CDF, Expectation | Medium | ||
| P-value definition | DS PDS AS | Statistics | P-value, Hypothesis testing | Easy | Google Meta Pinterest Snap | |
| Tests for normality | DS PDS AS | Statistics | Hypothesis testing, Normality | Easy | Duolingo Snap |