Technical practice
How would you structure a maintainable Python application that supports an AI feature?
What to coverThe posting calls for Python applications, reusable code, OOP, data structures and algorithms. Suggestion: Explain how you would separate data preparation, model interaction and application logic, and justify a data structure or algorithm choice.
How would you prepare and validate data before using it to train a model?
What to coverThe role includes preprocessing, cleaning, validation and transformation. Suggestion: Walk through checks for missing values, duplicates and inconsistent types, then explain how you would prevent training data from affecting evaluation results.
How would you train and test a basic machine-learning model?
What to coverThe posting asks for support with model training and testing and familiarity with Scikit-learn. Suggestion: Discuss a baseline, a suitable train/test split, a metric matched to the task and how you would investigate poor results.
How would you assess whether a prompt produces useful LLM responses?
What to coverLLMs, generative AI and prompt engineering appear under additional responsibilities. Suggestion: Describe task-specific examples, criteria for response quality and the trade-off between adding prompt detail and keeping prompts manageable.
How would you design and test a simple agentic workflow?
What to coverWorking knowledge of agentic AI and use of LangChain, LangGraph and MCP are listed requirements. Suggestion: Explain the roles of workflow steps and tools, when an explicit graph helps, and how you would check tool outputs and failure paths without assuming a particular project.
How would you expose a model through a backend service?
What to coverThe role includes integrating models with web applications or backend services and calls for REST API and JSON knowledge; Flask and FastAPI are awareness areas. Suggestion: Describe request validation, response format, error handling and what you would verify before deployment.
How would you troubleshoot a slow Python application that uses a model?
What to coverDebugging, troubleshooting and performance improvement are stated duties. Suggestion: Explain how you would measure time spent in data processing, model execution and service calls before choosing an optimization, and how you would confirm it preserves correct results.
When would you use SQL rather than Pandas during data preparation?
What to coverThe posting lists basic SQL and database knowledge alongside Pandas familiarity. Suggestion: Compare database-side filtering and aggregation with in-memory transformations, including data volume and ease of validation.
What would you check when a deep-learning model performs poorly?
What to coverDeep-learning basics and familiarity with TensorFlow or PyTorch appear under additional responsibilities. Suggestion: At a basic level, distinguish data problems from training problems and explain how you would use training and validation results to guide the next check.
How would you test and maintain a Python script used in an AI workflow?
What to coverDeveloping, testing and maintaining scripts, following coding standards, and documenting workflows are stated duties; Git or GitHub exposure is listed. Suggestion: Discuss tests for normal and invalid inputs, reviewable changes and documentation that helps someone reproduce the workflow.
Behavioural practice
Tell me about a time you resolved a difficult application defect.
What to coverDebugging and troubleshooting are part of the advertised work. Suggestion: Use situation, action and result to explain how you narrowed down the cause, verified the fix and observed its effect.
Describe a time you worked with others to integrate a model or data-dependent feature.
What to coverThe role includes assisting with model integration into web applications or backend services. Suggestion: Give a real situation, your specific actions to agree on inputs and outputs or resolve an integration issue, and the result.
Tell me about a proof of concept where you had to decide what to test first.
What to coverThe posting mentions contributions to proofs of concept and innovation initiatives. Suggestion: Describe the situation, the experiment and criteria you chose, what you did, and the result or limitation you reported.
Describe a time your documentation helped someone maintain or understand technical work.
What to coverDocumenting technical solutions, workflows and learnings is a stated responsibility. Suggestion: Use situation, action and result to show what you documented, how you made it usable and what happened afterward.