Design an End-To-End ML System Architecture

Homework system design: propose an architectural design and explain why you chose a specific tech stack and what factors you considered.

Asked in the Homework assignment stage. 2 system design questions in the take-home.

Prepare a practical design for an ML system of your choice. State assumptions before selecting components, and explain how the system would support data ingestion, training, evaluation, deployment, inference, monitoring, and iterative improvement.

Address the following:

1. Define functional and non-functional requirements.
2. Describe the offline data and training pipeline.
3. Design the online serving path, including retrieval, ranking, or re-ranking where applicable.
4. Justify the proposed technology stack and key alternatives.
5. Explain evaluation, monitoring, failure handling, feature drift, and training-serving skew.

Do not assume a particular business domain. Make clear which scale, latency, availability, cost, and compliance assumptions your design requires.