Interview prep
LLM system design answers
Worked answers to the design questions AI interviews ask, in the order you would say them in 35 minutes: clarify, size it, draw it, go deep, name the failures, prove it with evals.
- A coding agent that fixes small bugs“Design an agent that takes a bug ticket, finds the code, makes a fix, runs tests and opens a pull request.”An agent with repository tools in a sandbox, tests as the main check, budgets, and a person reviewing every pull request.
- Question answering over long legal contracts“Design a tool where a legal team uploads contracts and asks questions such as "what is the termination notice period?"”Long context with prompt caching for one contract, retrieval across many, strict citations to clauses, and lawyer review of risky answers.
- Email and ticket triage for a support team“Design a system that reads incoming support emails, classifies and prioritises them, routes them, and drafts replies.”An AI step with structured output, business rules on top, routing in the help desk, and drafts a person approves.
- Enterprise search with permissions“Design an AI search assistant over a company's SharePoint, Jira and Google Drive, where each employee sees only what they are allowed to.”Connectors that sync content and permissions, permission-aware retrieval inside the search, and grounded answers with citations.
- An evaluation pipeline for an LLM product“Design how a team would evaluate and monitor an LLM feature from first prototype to production.”A golden set and scorers offline, a regression gate in CI, and online evals with tracing and feedback that feed back into the set.
- Invoice extraction for an accounts team“Design a system that reads supplier invoices (PDF, photos, email) and posts them into the accounting system.”Document parsing, schema-based extraction, code validation of every field, and a review screen for failures.
- An internal LLM platform for many teams“Design a shared platform so 20 teams in a company can use LLMs safely, with control of cost and data.”A gateway with per-team keys and budgets, model aliases and fallbacks, masked logging, shared evals and approved models.
- A refund agent for an online store“Design an agent that handles refund requests end to end: checks the order, decides, and issues the refund.”A tool-using agent with task-shaped tools, rules for what it may do alone, human approval above a limit, and full audit.
- A customer support assistant for an Indian bank“Design an AI assistant that answers customer questions for a bank, in English and Hindi, on the website and WhatsApp.”RAG over approved policy documents, read-only account tools behind login, strict guardrails, and a clean hand-off to people.
- A voice agent that books appointments“Design a phone voice agent for a clinic chain that books, moves and cancels appointments in English, Hindi and Tamil.”Speech to text, an LLM with booking tools, text to speech, tight latency, and hand-off to the front desk.