← Google Interview Insights

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

Senior
Apr 2026

Summary

Got a system design question for an ML engineer role at Google. Pretty open-ended, which I wasn't fully prepared for.

Questions Asked (1)

Q1

Design a system that removes noise from audio signals.

System DesignTechnical Trade-offsAdaptability & Ambiguity
Author's notes

I started with a pretty basic spectral subtraction approach and the interviewer just kind of waited, so I figured they wanted more.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the requirements: what kind of noise, real-time vs offline, and quality metrics. Then propose a hybrid system combining classical signal processing (e.g., spectral subtraction) with deep learning models (e.g., U-Net on spectrograms), and discuss trade-offs in latency, accuracy, and computational cost.

Pro tip: Emphasize that noise removal is ill-posed without assumptions; explicitly state your assumptions about noise characteristics and signal type, and mention that you would evaluate with both objective metrics (e.g., SNR, PESQ) and subjective listening tests.

1. Clarify Requirements and Constraints

Ask questions to understand the use case: real-time or offline, type of noise (stationary vs non-stationary), acceptable latency, available compute, and target quality metrics.

2. Choose Architecture and Model

Decide between classical methods (spectral gating, Wiener filter) and deep learning (CNN, RNN, Transformer on spectrograms). Consider a hybrid approach for robustness.

3. Design Data Pipeline and Training

Plan how to obtain paired clean/noisy data (synthetic mixing or real recordings), augment data, and train with appropriate loss functions (e.g., L1 on spectrogram, SI-SNR).

4. Address Deployment and Scalability

Discuss model compression (quantization, pruning), serving infrastructure (batch vs streaming), and integration with existing audio pipelines.

5. Evaluate and Iterate

Define evaluation metrics (SNR, PESQ, STOI), set up A/B testing, and plan for continuous improvement with user feedback.

Key Points to Mention

  • Trade-offs between latency and accuracy: real-time systems may need lightweight models or block processing.
  • Choice of representation: time-domain (WaveNet, Conv-TasNet) vs frequency-domain (spectrogram masking).
  • Handling non-stationary noise: deep learning models can adapt, but require diverse training data.
  • Evaluation metrics: objective (SNR, PESQ, STOI) and subjective (MOS) measures.
  • Scalability: model size, inference speed, and cost for large-scale deployment.
  • Ethical considerations: potential bias in training data and impact on different accents or languages.

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