I have asked someone from OpenAI to do a System Design, here is what happened.

I run a discord server where we have a lot of engineers, especially senior and staff level and I decided to record a video with one of the most senior engineer we had as I was curious what it takes to

I run a discord server where we have a lot of engineers, especially senior and staff level and I decided to record a video with one of the most senior engineer we had as I was curious what it takes to clear the hardest loops in the industry. What he told us the expectations are for

Staff System Design Interview: You solve the hardest, most ambiguous problems with minimal input. In an interview, you drive the discussion from requirements gathering through to core system design and long-term considerations such as maintenance and future development. You should be able to describe the reliability, availability, and resource costs and trade-offs of your system, aligned with the design challenge. Furthermore, you should proactively cover cross-cutting concerns like operational and deployment toil, security, privacy, and team hiring. You can make reasonable decisions around build vs buy given budget constraints and internal control, and describe bottlenecks at multiple scales—from architectural choices to OS-level performance concerns—with minimal guidance. The problem I asked him was to design a Online Shopping Store like SHEIN. He picked the problem himself as he was very curious how they see a new dress trending online in just 2 weeks produce those items massively. This is a very untraditional problem, but allowed him to focus and go into a lot of depth. The design ideas are very different from what I would see from a typical mid level engineer, it included: The engineer dove deep into an architecture I’d never even considered, especially around: Execution in Milestones: Something that I have not seen before, on top of the class "Five Steps of System Design" (FRs, NFs, Calculations, API Design, Entity Design, HLD) - he added more steps to show seniority and technical leadership, specifically clarifying milestones of how things should be structured. Focus on the details rather than breadth: A lot of mid level folks try to come up with 10/15 requirements and execute on just 2-3. In his design, we saw that it is much more important to go into depth to delivery staff-grade system design performance. Really well-through through data design choices: In some projects, 80-85% of the challenge is the data design - sometimes, answering the question of how should the data be structured would simplify the problem significantly, he went well beyond just saying "User", "Order", "Account" That was really important. Event-Driven Microservices instead of simple CRUD: An event bus (e.g., Kafka) for everything from pricing changes, new product drops, to manufacturing updates, so that the system is highly reactive and can quickly coordinate supply, distribution, and marketing. APIs Beyond REST: 99.9% of people just use REST and never consider anything else. He favored gRPC internally to keep microservices fast and typed with Protobuf messages—then a BFF (Backend for Frontend) with REST for the mobile/web clients. Massive Data Infrastructure: To handle 400k new SKUs/year, you need a robust data pipeline (possibly with Cassandra or a distributed SQL (CRDB) store w/ hot-cold storage, plus Snowflake for analytics) that can ingest, transform, and store insights for quick lookups. I recommend going through the design/video to understand it, pretty great problem. Overall, this is one of the most impressive designs that I have ever seen. I think that most mid level folks should focus and try and replicate this. Exalidraw: <a href="https://app.excalidraw.com/s/17vCuvJeiD1/5K5DG0NuctQ)