Practice · carbon.talent
Machine Learning Engineer Interview Prep
Last updated 18 Oct 2026
Machine learning engineer interviews usually include coding, ML fundamentals, ML system design and questions about running models in production. Below are original practice questions with notes on what strong answers cover. These are original practice questions written for this page. They are not taken from, and do not preview, any assessment an employer or carbon.talent uses.
ML system design
- Design a system that recommends articles on a news site. Cover data, features, candidate generation, ranking, serving latency, evaluation and feedback loops.
- Design a fraud detection service for card payments that must respond within a strict latency budget. How do you handle class imbalance and changing fraud patterns?
Strong answers start with the objective and constraints, define offline and online metrics and spend time on monitoring and retraining, not only the model.
Production issues
- A model's accuracy dropped after a release, but offline metrics looked fine. How do you investigate?
- How would you prevent differences between features at training and serving time?
- How would you decide when to retrain a model?
Interviewers look for training-serving skew, data drift, shared feature definitions, canary releases and rollback plans.
Fundamentals
- Explain the bias-variance trade-off with an example from a project.
- When would you choose a gradient-boosted tree model over a neural network?
- How do you choose evaluation metrics for an imbalanced classification problem?
LLM evaluation
If the role involves large language models, expect questions such as: how would you evaluate a retrieval-based question-answering system? Cover a representative test set, retrieval and answer quality measures, human review, regression checks before release and cost and latency tracking.
Coding
Coding rounds are usually standard software problems in Python plus data manipulation. Practise writing clean, tested functions and explaining complexity.
Common questions
- What do ML engineer interviews cover?
- Coding, ML fundamentals, ML system design, production issues and, increasingly, LLM evaluation.
- How should I structure an ML system design answer?
- Objective and constraints, data and features, model approach, serving, offline and online evaluation, monitoring and retraining.
- Are these questions from carbon.talent's assessments?
- No. They are original practice questions and do not preview any employer or carbon.talent assessment.
- How do I prepare for production questions?
- Prepare examples of drift, skew and incidents you handled, and how you prevented recurrence.