AI Engineering Interview Questions — Practice Library | Velocode

This week's hardest·Reported at MetaNext Word Prediction System Architecture Design→

200+ problems

Problem

Tags

Type

Embedding Model Version Mismatch Between Development Production Environments

Reported at OpenAISWE (AI-Era)Easy

Code Comprehension

AI Agent Prompt Injection Attack Via Support Ticket Content

Reported at AnthropicSWE (AI-Era)Hard

Code Comprehension

Apache Airflow — Designing a Daily Pipeline with External Dependencies

Data EngineeringMedium

LLM Pipeline PII Redaction Architecture Design and Compliance Strategy

Reported at JP Morgan ChaseAI SecurityMedium

RAG evaluation metrics masking real-world answer quality failures

Reported at AnthropicSWE (AI-Era)Medium

Code Comprehension

Data Quality — Preventing Three Production Incidents

Data EngineeringHard

Silent Data Drift: Detecting Feature Distribution Changes Without Model Updates

Reported at DatabricksSWE (AI-Era)Medium

Code Comprehension

Silent data drift detection in production ML pipeline accuracy collapse

Reported at DatabricksSWE (AI-Era)Medium

Code Comprehension

Deployment Patterns — Canary Rollout for a New Recommendation Model

ML EngineeringMedium

Streaming Clickstream Lakehouse: Storage, Compaction, Queries, Schema Evolution

Reported at DatabricksData EngineeringHard

Reranker degrades pipeline performance despite correct isolation testing behavior.

Reported at CohereSWE (AI-Era)Medium

Code Comprehension

Privacy in ML — GDPR Erasure Requests for a Healthcare Model

AI SecurityHard

Building Customer Support Agent Evaluation Harness: Metrics Design and Contamination Prevention

Reported at AnthropicAI EngineeringMedium

Users with session count growth for three consecutive months

Reported at AirbnbData EngineeringMedium

Spark Internals — Diagnosing and Fixing a Skewed Join

Data EngineeringHard

Model Context Protocol standardizes AI tool integration architecture

Reported at AnthropicAI EngineeringMedium

Spark Join OOM: Check broadcast threshold and shuffle partition configuration

Reported at DatabricksData EngineeringHard

Model Monitoring — Diagnosing Drift in a Credit Scoring Model

ML EngineeringHard

RAG Chatbot Fails to Answer Despite Successfully Retrieving Relevant Documents

Reported at GoogleSWE (AI-Era)Easy

Code Comprehension

Infinite Loop Detection: Root Cause Analysis and Design Prevention

Reported at AnthropicAI EngineeringMedium

Model Serving — Batch vs Real-Time Architecture Decision

ML EngineeringMedium

Delta Lake Time Travel: Querying Historical Table Versions

Reported at DatabricksData EngineeringHard

Agent hallucinating confident incorrect data from tool composition failures

Reported at OpenAISWE (AI-Era)Hard

Code Comprehension

LLM Summarizer Producing Hallucinated Content At High Failure Rate

SWE (AI-Era)Medium

Code Comprehension

Loading more problems…