Interview questions
Uber AI engineer interview questions (2026)
7 real interview questions reported by engineers who interviewed at Uber, spanning Data Engineering, ML Engineering, AI Engineering. Every question is scored against a golden answer on the things Uber actually grades, architecture, token efficiency, security and correctness, not just whether your code runs.
Uber ML Engineering questions
- Feature Store — Backfilling a New Feature Without Downtime
A data scientist adds a new feature to an existing model. You must make it available for training with full history and low-latency serving, without downtime and without corrupting
- Feature Store — Eliminating Online/Offline Skew for Real-Time Fraud
You are building real-time fraud features like transactions-per-card-last-hour and average-amount-last-30-days, served under 50ms. Design the feature store: offline vs online split
- Feature Store — When It Earns Its Complexity
A team argues a feature store is over-engineering and they should just write feature SQL per model. Give the decision framework: what concrete problems a feature store solves that
Uber Data Engineering questions
- Ride-level fact table with trip completion metrics and surge multipliers
Design a schema for a ride-sharing fact table that supports surge pricing analytics. What grain do you pick and why?
- User lifetime net revenue and distinct active months count
Given two tables, write a single SQL query returning each user's lifetime net revenue and their number of distinct active months. ## Input - `orders(order_id INTEGER, user_id INT
Uber AI Engineering questions
- Pixel-level classification assigning class labels to image regions
Explain semantic segmentation.
- Instance segmentation architecture combining region proposals with convolutional neural networks
Explain Mask R-CNN.
Prepare for your Uber interview
Velocode turns reported Uber interview questions into scored practice. Free accounts get the full community problem set and one Uber-tagged question per domain; Pro unlocks every company-verified question, the interview simulator, and your domain readiness radar.