Generative AI & LLMs
GenAI System Design
11 practice questions. Free questions open a full answer guide; the rest unlock with Pro.
- Walk me through how you'd design a question-answering feature over our internal documents using an LLM. What are the main pieces?
- Your company wants to start using AI agents to automate real internal workflows. As the engineering manager, how do you decide where to apply them, and how do you lead the rollout so an autonomous system taking real actions doesn't become a liability?Go Pro
- You're designing a knowledge-grounded assistant over a large, frequently-changing corpus of internal documents, and the team is split between stuffing everything into a long context window, building retrieval, and fine-tuning the model on the docs. How do you reason through that architecture, and how do you keep the answers grounded?Go Pro
- An LLM feature is getting popular and the API bill is climbing fast. How would you bring the cost down without obviously hurting quality?Go Pro
- A consumer LLM feature is gaining traction but the per-request model bill and latency are climbing faster than revenue. How would you redesign the system to bring cost and latency down while keeping answer quality acceptable?Go Pro
- You're designing the architecture for a production LLM feature that has to meet a latency SLA and stay within budget at scale. How would you structure the system around the model call?Go Pro
- You're adding streaming to a production LLM feature so users see tokens as they're generated. How would you design the streaming path, and what does streaming make harder — moderation, error handling, and cancellation?Go Pro
- Before launching an LLM-powered answer feature over our help-center content, how would you tell whether it's actually good enough to ship?Go Pro
- A prototype LLM feature works great in a notebook, and leadership wants it in production next quarter. As the senior engineer, how do you take it from a working demo to a reliable, operable production service?Go Pro
- Your product depends on a third-party LLM API. What happens to your feature when that provider is slow, rate-limits you, or goes down, and how would you design so users aren't stuck?Go Pro
- Design a customer-support agent that doesn't just answer questions but takes real actions on a user's account — issuing refunds, changing a subscription, canceling an order. Walk me through the architecture, focusing on how you keep those actions safe.Go Pro
Want questions matched to your role? Paste a job title, job description, or CV for a personalized set, or go Pro to unlock the full bank.