Design a streaming embedding-based classifier
Company: Apple
Role: Machine Learning Engineer
Category: ML System Design
Difficulty: hard
Interview Round: Technical Screen
You are given a continuously arriving stream of text data for a classification task. Design an end-to-end machine learning system that:
1. processes raw text online,
2. tokenizes the text,
3. converts tokens into embeddings,
4. trains a classification model, and
5. serves low-latency predictions in production.
Explain your choices for data preprocessing, tokenization, embedding generation, model architecture, training strategy, evaluation metrics, and deployment. Also discuss how you would handle large data volume, model updates, and consistency between training and serving.
Overview: This question evaluates a candidate's ability to design end-to-end streaming machine learning systems, including online text preprocessing, tokenization, embedding generation, continuous model training, and low-latency classification serving.