Your question is Handle Bad Records in Streaming ML. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
You're working on a real-time machine learning pipeline where records arrive continuously from operational databases and application events. Some records are missing required fields, while others are corrupted, duplicated, or malformed. You need a practical approach for keeping the pipeline usable without letting bad data silently degrade downstream features or predictions.
How would you handle missing or corrupted database records in a real-time machine learning pipeline?