Machine Learning interview questions

Showing 91–120 of 572 results

• Build a leak-free sklearn pipelineMedium

  This problem tests the ability to build a robust scikit-learn pipeline that prevents data leakage during preprocessing and hyperparameter tuning while handling class imbalance.

  Boston Consulting GroupData ScientistMachine LearningOnline Assessment
• Build and evaluate imbalanced binary classifierMedium

  This question evaluates a data scientist's skills in building reproducible machine learning pipelines for imbalanced binary classification, covering temporal splitting to avoid leakage, class imbalance handling, feature preprocessing, probability calibration, threshold selection, and monitoring for calibration drift.

  Boston Consulting GroupData ScientistMachine LearningOnline Assessment
• Design and sample for credit default predictionMedium

  This question evaluates a data scientist's competencies in imbalanced binary classification, temporal label definition and leakage detection, sampling and calibration under prior-probability shift, metric and operating-point selection, cost-sensitive decision rules, time-based validation and monitoring, and fairness and operational guardrails.

  Boston Consulting GroupData ScientistMachine LearningTechnical Screen
• Detect Data Leakage in Supervised Learning PipelinesHard

  This interview question evaluates core ML concepts, assumptions, math intuition, training/evaluation trade-offs, and practical failure modes in a realistic interview setting. A strong answer for Detect Data Leakage in Supervised Learning Pipelines states assumptions, handles edge cases, explains trade-offs, and shows how to validate the result clearly.

  Boston Consulting GroupData ScientistMachine LearningOnline Assessment
• Improve Model Generalization with Cross-Validation and Feature EngineeringMedium

  This interview question evaluates core ML concepts, assumptions, math intuition, training/evaluation trade-offs, and practical failure modes in a realistic interview setting. A strong answer for Improve Model Generalization with Cross-Validation and Feature Engineering states assumptions, handles edge cases, explains trade-offs, and shows how to validate the result clearly.

  Boston Consulting GroupData ScientistMachine LearningOnline Assessment
• Reduce overfitting under constraintsHard

  This question evaluates a candidate's competency in machine learning for mitigating overfitting in tabular regression under production latency constraints, testing knowledge of regularization, model complexity control, feature engineering, validation strategies, and experimental design.

  Boston Consulting GroupData ScientistMachine LearningTechnical Screen
• Scale and Normalize: When to Use Each Method?Easy

  This interview question evaluates core ML concepts, assumptions, math intuition, training/evaluation trade-offs, and practical failure modes in a realistic interview setting. A strong answer for Scale and Normalize: When to Use Each Method? states assumptions, handles edge cases, explains trade-offs, and shows how to validate the result clearly.

  Boston Consulting GroupData ScientistMachine LearningOnline Assessment
• Train GradientBoostingClassifier with 5-Fold Cross-ValidationEasy

  This interview question evaluates core ML concepts, assumptions, math intuition, training/evaluation trade-offs, and practical failure modes in a realistic interview setting. A strong answer for Train GradientBoostingClassifier with 5-Fold Cross-Validation states assumptions, handles edge cases, explains trade-offs, and shows how to validate the result clearly.

  Boston Consulting GroupData ScientistMachine LearningOnline Assessment
• Choose Precision, Recall, or F1 for Agent EvaluationMedium

  This question tests the ability to map business objectives and error costs to appropriate classification metrics in a real-world system context.

  ByteDanceSoftware EngineerMachine LearningTechnical Screen
• Choose Ranking Functions, Customer Value Metrics, and Predictive ModelsMedium

  This problem tests the ability to implement SQL window functions for cohort analysis, define robust customer lifetime value metrics, and articulate the trade-offs between various machine learning algorithms in a production setting.

  ByteDanceData ScientistMachine LearningTechnical Screen
• Compute Sentence SimilarityMedium

  This question evaluates understanding of sentence-level and token-level embedding techniques, text preprocessing, similarity metrics, edge-case handling for empty or unknown tokens, and trade-offs between pretrained sentence encoders and averaged word embeddings.

  ByteDanceMachine Learning EngineerMachine LearningTechnical Screen
• Explain GRPO for LLM Reinforcement Learning and How It Differs from PPOMedium

  An ML question asking you to explain Group Relative Policy Optimization (GRPO) for fine-tuning large language models with reinforcement learning. It tests your grasp of group-normalized advantages, the clipped objective with a KL penalty, the contrast with PPO and its value model, and common failure modes.

  ByteDanceApplied ScientistMachine LearningTechnical Screen
• Explain LLM generational improvements, multimodal LLMs, and normalization in LLMsMedium

  Three conceptual LLM questions: the main improvements across generations of a model family as described in technical reports, how multimodal image-and-text models are built and trained, and why modern LLMs use layer normalization or RMSNorm with pre-norm placement instead of batch normalization.

  ByteDanceMachine Learning EngineerMachine LearningTechnical Screen
• Explain XGBoost's Overfitting ResistanceMedium

  This question evaluates understanding of ensemble learning and model generalization, focusing on why gradient-boosted tree ensembles are often less prone to overfitting than single decision trees.

  ByteDanceMachine Learning EngineerMachine LearningTechnical Screen
• Hand-Write a Multi-Head Attention Layer: Projections, Head Splitting and MaskingMedium

  Write a multi-head attention layer from scratch: project queries, keys and values, split them into heads, apply scaled dot-product attention with optional padding or causal masks, and merge the heads. It tests tensor shape discipline, numerical stability, and the cost of attention as sequence length grows.

  ByteDanceSoftware EngineerMachine LearningOnsite
• How to deploy and tune multimodal models?Hard

  A six-part ByteDance data scientist onsite ML interview covering deploying and tuning multimodal models under GPU/VRAM, compute, and latency constraints; designing fast video retrieval over precomputed captions and embeddings with ANN and hybrid reranking; and deep-learning fundamentals — overfitting detection and mitigation, dropout and inverted-dropout scaling, normalization layers (BatchNorm/LayerNorm/GroupNorm/InstanceNorm/RMSNorm), and reinforcement learning for LLM post-training (RLHF). It tests practical systems tradeoffs alongside theoretical foundations across ML systems, retrieval, and reinforcement learning.

  ByteDanceData ScientistMachine LearningOnsite
• How would you choose a classification threshold?Medium

  This question evaluates understanding of threshold selection for probabilistic binary classifiers, testing competencies in validation-set decision threshold choice, cost-sensitive trade-offs between false positives and false negatives, effects of class imbalance, and probability calibration.

  ByteDanceData ScientistMachine LearningTechnical Screen
• Implement Multi-Head Attention from ScratchMedium

  An ML coding question asking you to implement multi-head attention from scratch in NumPy or PyTorch, from the query, key and value projections to the output projection. It tests tensor shape handling, scaled dot-product attention, masking, numerical stability and complexity analysis.

  ByteDanceApplied ScientistMachine LearningTechnical Screen
• Implement a Transformer Decoder from Scratch Without Starter CodeHard

  Implement a transformer decoder from scratch with no starter code in about 30 minutes, including embeddings, masked multi-head self-attention, feed-forward layers, residual connections and layer normalization. Tests tensor-shape fluency, causal masking and hands-on deep learning coding for a machine learning engineer interview.

  ByteDanceMachine Learning EngineerMachine LearningTechnical Screen
• Predict Short-Video Engagement with Skewed DataHard

  Design a short-video engagement model with explicit targets, leakage-safe features, heavy-tail handling, exposure awareness, and temporal evaluation.

  ByteDanceData ScientistMachine LearningTechnical Screen
• Write the Cross-Entropy Loss and Compute Precision, Recall and F1Medium

  An ML fundamentals question in two parts: write the cross-entropy loss for binary and multi-class classification and explain why it is used, then compute precision, recall, F1 and accuracy for a described imbalanced classifier. It tests exact formulas, careful arithmetic and metric choice.

  ByteDanceApplied ScientistMachine LearningTechnical Screen
• Choose Tokenization and Dense or Sparse RetrievalMedium

  This problem tests your understanding of document retrieval strategies, specifically comparing sparse and dense representations and how to integrate them into a production RAG system.

  C3 AIData ScientistMachine LearningTechnical Screen
• Choose a Parameter to Reduce Underfitting in Gradient-Boosted TreesMedium

  Explain why decreasing minimum observations per leaf can reduce gradient-boosted tree underfitting and how to validate the change.

  C3 AIData ScientistMachine LearningOnline Assessment
• Compare How Models Respond to MulticollinearityMedium

  Compare multicollinearity effects in regression, Lasso, logistic regression, decision trees, and boosted trees without overstating immunity.

  C3 AIData ScientistMachine LearningOnline Assessment
• Compare PCA and Autoencoders Under the Curse of DimensionalityHard

  This question tests your conceptual understanding of dimensionality reduction techniques and your ability to weigh their performance in high-dimensional feature spaces prone to the curse of dimensionality.

  C3 AIData ScientistMachine LearningTechnical Screen
• Compare Transformers and RNNs and Derive Attention ComplexityHard

  Evaluate candidate's deep learning foundations by comparing Transformer and recurrent architectures and deriving exact time/memory complexities for standard scaled dot-product attention across training and inference.

  C3 AIData ScientistMachine LearningTechnical Screen
• Connect Bias-Variance Trade-offs to Bagging and BoostingHard

  This question tests the conceptual understanding of how ensemble methods manipulate model error components through different structural paradigms.

  C3 AIData ScientistMachine LearningTechnical Screen
• Evaluate Factual Accuracy and Consistency in a Domain-Specific RAG ChatbotMedium

  Evaluate a rare-disease RAG chatbot using domain consistency and expert review, and distinguish factual accuracy from retrieval metrics and lexical overlap.

  C3 AISoftware EngineerMachine LearningOnline Assessment
• Explain key ML/stats conceptsMedium

  Explain key ML/stats concepts evaluates core ML concepts, assumptions, math intuition, training/evaluation trade-offs, and practical failure modes in a realistic interview setting. A strong answer states assumptions, handles edge cases, explains trade-offs, and shows how to validate the result clearly.

  C3 AIData ScientistMachine LearningOnsite
• Explain the Margin View of Boosting ClassificationMedium

  Understand boosting classification through sequential learners, signed margins, surrogate losses, and the limits of training-margin claims.

  C3 AIData ScientistMachine LearningOnline Assessment