System Design - Design an Autonomous Vehicle Ride-Hailing App (Waymo-like)

This problem statement is designed for a Senior Data Engineer or Solutions Architect interview at Databricks. It evaluates a candidate's ability to design a Lakehouse architecture that handles high-velocity IoT streams, complex real-time matching, and large-scale analytical reporting.

Interview Problem: Waymo-style Autonomous Ride-Hailing Platform

Background: You are tasked with designing the data backbone for "AutoRide," a fleet of fully autonomous vehicles. The system must process massive streams of vehicle telemetry, handle real-time ride requests, and provide executive dashboards for fleet health.

1. Data Requirements & Scale

• Vehicle Telemetry: 50,000 autonomous vehicles emitting GPS, speed, and sensor health every 5 seconds.
• Ride Requests: 200,000 requests per hour during peak times.
• Latency: Ride matching must occur in under 1 minute.
• Retention: Raw sensor logs must be kept for 7 years for regulatory and safety audits.

2. Core Functional Requirements

• Real-Time Fleet Monitoring: Live dashboard showing active vehicles, current battery/fuel levels, and maintenance alerts.
• Ride Matching Engine: A service to find the optimal vehicle for a rider based on current proximity and estimated time of arrival (ETA).
• Safety Incident Analysis: Ability to "time travel" to reconstruct vehicle state exactly as it was during a reported safety disengagement.
• Dynamic Pricing: Hourly updates to ride costs based on regional demand and fleet availability.

3. Technical Design Tasks (Candidate Expectations)

1. Ingestion Strategy: Propose how to ingest high-frequency IoT data into the Databricks Lakehouse using Auto Loader or Structured Streaming.
2. Medallion Architecture: Define the schemas for Bronze (raw logs), Silver (cleaned telemetry and ride events), and Gold (fleet performance metrics) layers.
3. Data Modeling: Design the table relationships between Vehicles, Riders, Trips, and Location_Pings.
4. Optimization: Explain how to use Z-Ordering or partitioning to ensure fast spatial queries (finding cars near a user).
5. Governance & Privacy: Describe how Unity Catalog would be used to mask PII (Personally Identifiable Information) like rider home addresses.

4. Example Follow-up Questions

• "How would you handle data skew if 80% of your rides are concentrated in San Francisco while the rest of the fleet is idle?"
• "A sensor update fails for 10% of the fleet during a software rollout. How do you design a re-run or recovery logic in your Databricks Workflow?"
• "How does Delta Lake’s Time Travel feature specifically solve the 'Safety Incident Analysis' requirement?"

Would you like a sample solution for the Medallion Architecture or a specific PySpark code snippet for the ride-matching logic?

[0] - Design a Ride-Sharing Service Like Uber [1] - Top 50 Databricks Data Engineering Interview Questions and ... [2] - 50 Databricks Interview Questions & Answers: The Ultimate ... [3] - Design Uber - System Design Interview Question [4] - How do you crack Data Engineering Interviews at Uber, Atlassian & ... [5] - Uber Data Engineer Interview: Design a Ride Sharing Schema [6] - Design Uber/Lyft (Part 1) | Systems Design Questions 3.0 With ... [7] - Ace the System Design Interview — Uber/Lyft - Medium [8] - Designing a Ride Sharing App like Uber: System ... - LinkedIn [9] - What to Expect in the Databricks Technical Assessment ... [10] - Databricks Data Engineer Interview Questions (Updated 2026)