AI0-001 AI Infrastructure and Technologies Practice Question
A platform team is preparing a feature store for a recommendation system. They need point-in-time correct feature retrieval so that training datasets do not leak future information, and they need the same features served online with low latency. Which architecture best satisfies both requirements?
⚠ Common exam trap
The trap here is assuming that one storage system can serve both batch training and low-latency online inference, when the two access patterns require different stores synchronized through the same transformation logic.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Maintain an offline store with time-stamped feature values for training and a synchronized online store keyed by entity for low-latency serving.
A feature store that pairs a time-stamped offline store with a synchronized online store supports point-in-time correct training joins and fast online retrieval from the same feature definitions. This dual-store pattern prevents label leakage and reduces training-serving skew. Single-store, lake-query, and duplicated-code approaches each fail at least one of correctness or latency.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Use a single relational database table that overwrites feature values in place with no timestamps.
Why it's wrong here
Overwriting values in place destroys the history needed for point-in-time correct training joins, because past feature states are lost. A single table without timestamps cannot reconstruct what the model would have seen at a past prediction time. This design causes label leakage and is unsuitable for feature management.
- ✗
Compute features on the fly in the training pipeline and again independently in the serving path using separate code.
Why it's wrong here
Separately implemented training and serving feature logic commonly causes training-serving skew, where the model sees different feature distributions in production than during training. It also lacks a historical store for point-in-time joins, risking label leakage. This approach undermines both correctness and consistency.
- ✓
Maintain an offline store with time-stamped feature values for training and a synchronized online store keyed by entity for low-latency serving.
Why this is correct
A dual-store feature platform keeps historical, time-stamped values in an offline store for point-in-time correct training joins, and materializes the latest values into a low-latency online store keyed by entity ID. This prevents label leakage during training while enabling fast online retrieval. It is the standard architecture for consistent offline and online features.
- ✗
Store features only in a data lake and have the online service query the lake on each request.
Why it's wrong here
Querying a data lake per request yields high latency unsuitable for online serving and provides no point-in-time correctness mechanism for training. Data lakes are optimized for batch analytics, not millisecond feature retrieval. This architecture fails both the online latency and training correctness requirements.
About these practice questions
One of 962 original AI0-001 practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official CompTIA exam blueprint
This AI0-001 practice question is part of Courseiva's free CompTIA certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI0-001 exam.