Snapchat Machine Learning Engineer Interview Experience — Model Deep Dives and Organization-Chart Coding

The author describes Snapchat onsite sessions divided between a short values-and-project discussion and approximately 50 minutes on the round’s technical subject. Values questions emphasized AI applications, prioritization, and empathy. The machine-learning discussion examined project decisions and training problems, including normalization, regularization, loss functions, and different prediction targets. The author found this questioning difficult and needed prompting to discuss neural-network regularization.

The coding exercise involved representing employee reporting relationships, displaying a hierarchy, and identifying relationships two levels apart. Its instructions allowed AI assistance while retaining responsibility for explanation, testing, edge cases, and complexity. Applied machine-learning and system-design sessions used advertising or recommendation scenarios. Follow-ups examined data, features, sampling, model choices, tuning, and the meaning and tradeoffs of evaluation metrics. The report does not state an interview outcome.