The take-home challenge was an open-ended machine learning task focused on exploring perplexity as a metric. There wasn’t much guidance, but the process felt fair overall. I gave it a strong effort, uncovered some useful insights, and made reasonable progress given the scope. Unfortunately, I wasn’t selected for the next stage. While the feedback and communication could’ve been clearer, I still appreciated the opportunity to tackle a real-world-style problem.