Google ML SWE Interview – L3 (2020)

Interview Outcome

Status: Offer received
Level: L3 – Machine Learning Software Engineer
Location: Google MTV
Date: July 2020

Candidate Background

• Master’s degree in Computer Science from a top 30 U.S. university

• 1.5 years of experience as an ML Engineer/Data Scientist

• Completed 2 research internships in Machine Learning
  
  

Interview Journey Overview

The candidate began the interview process in 2019, starting with two phone interviews followed by an onsite round (pre-COVID, in-person). Despite receiving 2 strong accepts and 3 accepts, Google's bar of 3 strong accepts meant the candidate was asked to attend two more ML-focused phone interviews. With those resulting in standard accepts again, the recruiter suggested another onsite after 8 months. The candidate chose to take more time and returned after 14 months — this time fully prepared.

Preparation Strategy

Technical Preparation

Coding Practice (LeetCode)

• Solved 456 LeetCode problems:

  • Easy: 26%

  • Medium: 54%

  • Hard: 20%

• Focused on “Top 100 Liked” and “Top Interview” questions initially

• Later used LeetCode Premium to target Google-tagged problems

• Adopted time-boxed practice: 30-minute attempts, then hints, then solutions

• Maintained a “To Review” list for important and frequently asked problems

• Prioritized reviewing over solving new problems in the final weeks
  
  

Machine Learning Knowledge

• Reviewed Andrew Ng's Deep Learning Specialization (Coursera)

• Studied ML class notes and cheat sheets like Scikit-Learn ML Map

• Took notes and prepared summaries for quick revision
  
  

Mental Preparation

• Adopted a consistent, small-step daily routine

• Dedicated 1–2 hours per day over several months

• Engaged in daily prayer and mindfulness to stay mentally focused

• Approached the interview as a learning journey, not just a test

• Remained calm and motivated by looking at interviews as personal growth milestones
  
  

Final Weeks Before Interview

• Focused on reviewing notes and questions

• Practiced mock interviews and behavioral answers

• Took rest in the last two days to be mentally refreshed

• Booked a quiet hotel room for the interview to avoid home distractions

• Used noise-cancelling AirPods to ensure focus during the session
  
  

Interview Breakdown

Machine Learning Rounds

• Asked to describe current ML role and responsibilities

• Warm-up: definitions of supervised vs. unsupervised learning

• Designed a deep learning network for an unlabeled task, including:

  • Choosing architecture

  • Selecting loss functions for different stages

  • Adapting the design if labels were available

• Knowledge areas assessed:

  • DL architectures

  • Training optimization

  • Real-world experience with deployed ML systems
    
    

Behavioral Round

• Covered conflict resolution with team members

• Handling credit disputes in collaborative projects

• Demonstrated leadership and initiative in complex situations

• Strategies for working with senior teams and cross-functional projects

• Used STAR format and prepared responses for common behavioral scenarios
  
  

Coding Rounds

Round 1 – Math & DFS

• Changing base of number representation

• Group permutations based on constraints

• DFS-based problem (non-obvious at first glance)

• Bit manipulation: computing bit requirements for a range
  
  

Round 2 – Arrays & Sliding Window

• Largest contiguous subarray satisfying a condition

• Sliding window-based logic

• Experienced a few difficulties but navigated through with composure

LeetCode and especially the review process played a crucial role in acing these rounds.

ML System & Job Experience Round

• Focused deeply on the candidate’s current job role

• Questions included:

  • Types of models used and rationale

  • Alternatives and why they weren't used

  • Data sources, refresh frequency, and evaluation methods

  • Model retraining triggers and update strategies

  • Impact of COVID-19 on model performance and adaptations
    

This round was unexpected but pivotal. Real job experience and end-to-end understanding made a strong impact.

Final Advice

• Small but consistent steps matter more than cramming

• Anyone can clear the interview with persistence and planning

• Treat interviews as learning opportunities, not just evaluations

• Focus on both technical excellence and mental resilience

• Stay positive — the results will follow eventually
  
  

“The journey matters as much as the destination. Keep learning, keep building, and keep your mind and soul strong.”