Voleon Machine Learning Engineer Interview Experience — A Time-Series Pipeline Coding Screen

After recruiter outreach, the author agreed to interview with Voleon and reports being rejected after a 90-minute machine-learning coding session. The exercise involved a pipeline with roughly five tasks and a CSV dataset containing time-series observations. The author remembers debugging, changing model parameters, running results, and extending the pipeline for inference, but not the remaining tasks.

The author suspected that feature construction introduced data leakage; the report does not establish that diagnosis conclusively. Although employed as a machine-learning engineer, the author describes doing mostly software engineering work and having little recent pandas practice. The author consulted pandas documentation during the session. The author recommends practical data-science exercises for preparation rather than relying entirely on algorithm drills.