← Waymo Interview Insights

Waymo·Machine Learning Engineer·Onsite - System Design / Architecture·Senior

Senior
May 2026

Summary

Waymo ML engineer interview with a system design question around video retrieval for autonomous driving scenarios. Pretty focused and technical, no fluff.

Questions Asked (1)

Q1

You're given a video clip of a driving scene (like a near-collision or heavy rain). Design a system that retrieves similar scenes from a massive video dataset.

System DesignTechnical Trade-offs
Author's notes

This is the kind of question where you can go a hundred directions and that's exactly what tripped me up at first.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying requirements: what defines 'similar' (e.g., near-collision, heavy rain), latency needs, and dataset scale. Then propose a two-stage retrieval system: first, extract compact embeddings from video clips using a video encoder; second, use approximate nearest neighbor search to retrieve top-k similar clips, followed by re-ranking if needed.

Pro tip: Emphasize the importance of a well-defined similarity metric and the trade-off between embedding dimensionality and retrieval speed; also mention the need for a robust evaluation protocol using human judgments or proxy tasks.

1. Clarify Requirements

Ask about the definition of 'similar', dataset size, latency constraints, and whether the system should handle multimodal queries (e.g., text + video).

2. Design Embedding Model

Choose a video encoder (e.g., 3D CNN, Transformer) to map clips to fixed-dimensional vectors; consider self-supervised pretraining on driving data for better representations.

3. Indexing and Search

Use approximate nearest neighbor (ANN) libraries (e.g., FAISS, ScaNN) to index embeddings for fast retrieval; discuss trade-offs between index size, speed, and recall.

4. Re-ranking and Refinement

Optionally re-rank top candidates with a more expensive model (e.g., cross-attention) to improve precision, especially for subtle similarities.

5. Evaluation and Iteration

Define metrics (e.g., recall@k, mAP) and create a validation set with human labels; iterate on model and index parameters based on performance.

Key Points to Mention

  • Choice of video representation: frame-level vs. clip-level features, temporal modeling
  • Similarity metric: cosine similarity, Euclidean distance, or learned metric
  • Scalability: distributed indexing, sharding, and handling billions of clips
  • Latency vs. accuracy trade-off: two-stage retrieval with ANN and re-ranking
  • Handling multimodal queries: combining visual, textual, and sensor data
  • Evaluation challenges: defining ground truth for 'similar' driving scenes

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.