How do you start a project for an assistant serving 1,500 employees?
Review note: The answer refers to the original author's CV or target job offer — adapt it to your own experience.
Answer out loud or in your head first, then compare.
Answer
First I choose one specific process: for example, answering questions about internal policies. I establish who uses it, which documents are needed, how quickly permissions change and whether the assistant only reads or also writes something. The number of accounts isn't enough to estimate traffic.
The first version can have company login, a simple interface, a backend that retrieves accessible documents and a model that creates an answer with sources. Updating the documents runs separately. Longer tasks can go to a queue. Only the requirements justify choosing specific AWS services.
I start with a pilot and tasks for evaluation. I measure correctness, time and cost. I decide what the system does when there's no source or the model fails. This is a proposed architecture, not a description of a confirmed RAS system.
Why it matters: Shows the order of thinking, from the problem to the components.
Possible follow-ups: What's the simplest scope that shows value? Where do you check access to data?
How did it go? Be honest — “Not yet” puts the question on your review list.
Start with
- D-P15 · What is idempotency? — Backend, data and system design
- G-P01 · What is RAG? — RAG and retrieval
- H-P01 · How does an AI agent differ from a plain model call? — Agents and workflows
Practise in code
- An assistant with a new requirement — guided self-review
Further reading
- Official job offer: Senior AI Engineer, RAS Tech(opens in a new tab)Job offer; accessed 22.10.2026. The full scope of the role, stack and integrations. Doesn't describe the exact interview task.
- Azure AI Search: Document-level access control(opens in a new tab)Docs; accessed 22.10.2026. Security filters, native ACLs and preview features; syncing permissions needs an explicit mechanism.