Machine Learning System Design Interview, issue 19, Dec 5, 2025

The Database-as-Queue Trap

Machine Learning Engineer interview at Google, and the interviewer asks:

We have 3 upstream microservices generating features. They write to a central 𝘗𝘰𝘴𝘵𝘨𝘳𝘦𝘴 𝘋𝘉. Your ML Service queries that DB to get the input vector for inference. How do we scale this to 50k requests per second?

How relying on a persistence layer for real-time inference creates silent bottlenecks, and how event brokers fix them.

The full answer, with the mechanism and the arithmetic, is for paid subscribers on Substack.

Read it on Substack

Get the next one

Free on Substack. Unsubscribe in one click.