Loading…
Coding interview questions with attempt-first practice, a hint ladder and step-by-step visual solutions: arrays, trees, graphs, DP and LLM problems.
34 questions
Not sure where to start? Follow the Study Plan — week by week, week 1 free.
Coding and system design are only two of the rounds
Interactive guides to the most common behavioral questions — decode the question, build a STAR structure, survive the follow-ups. Free.
TierOnePrep offers a set of fully unlocked free questions — coding problems spanning easy, medium, and hard (including a binary tree walkthrough with the tree drawn at every step), plus interactive system design and low-level design guides — with no account required. The question of the day is also unlocked for everyone, whichever question it lands on. The rest of the bank and its guides are available on the Pro plan.
Each interactive system design guide walks through a 4-phase interview simulation: Brief (the vague interviewer prompt), Clarify (must-ask and good-to-ask questions), Design (architecture, diagrams, calculations, and trade-offs), and Follow-ups (difficulty-rated questions with hints).
System design questions ask you to build a system from primitives and are assessed largely on the capacity arithmetic showing the design survives its own load. Cloud architecture questions assume the building blocks exist and assess service selection instead: self-managed versus managed trade-offs, migration sequencing, failure behaviour and cost. Both render the same interactive four-phase walkthrough, but a cloud architecture guide replaces the back-of-envelope calculations with a managed-versus-self-run service comparison. Filter by type on this page to see either set on its own.
Also in the loop, free and outside the bank: Behavioral (10) and Discussion round (12).
A low-level design question stays inside a single process, and this bank holds two kinds. The object-oriented ones — also called OOD or the machine coding round — ask how the code is organised: designing a parking lot, an elevator, a vending machine or a chess game, judged on class responsibilities and which interfaces absorb the requirements that will change. The operational ones are the same round as asked of SRE, DevOps and platform engineers, with a component or a workflow as the unit of design and no code: a CI/CD pipeline, a rollout and rollback controller, health checking, secrets rotation, configuration distribution, backup and restore. System design asks how a system spans machines and is judged on whether it survives its own load; low-level design of either kind is judged on whether it survives the next requirement. Every card is labelled "Includes code" or "No code", and the two sets have separate topic pages.
A troubleshooting question gives you a symptom instead of a requirement: customers report that login fails sometimes, p99 latency tripled after a deploy while p50 did not move, a nightly job reports success and produces nothing. The interviewer holds a specific cause and reveals it only in answer to your questions, so the round is a test of method rather than of knowledge. It is scored on whether you establish the blast radius before forming a theory - all users or some, since when, correlated with what change - whether each step you take rules something out, and whether you separate stopping the damage from finding the cause. Each guide works the question the way the round runs: the clarifying questions to ask and what each answer eliminates, a top-down narrowing ladder, and the causes the symptom is genuinely consistent with, with the fingerprint that tells them apart. These questions are standard in SRE, DevOps, platform and cloud loops and increasingly in senior software engineering loops.
Yes. Questions are modeled on real interview patterns at FAANG companies (Facebook/Meta, Apple, Amazon, Netflix, Google) and AI-first companies like Anthropic, OpenAI, Nvidia, and SpaceX — including AI-infrastructure topics like LLM inference and GPU cluster scheduling that come up at Anthropic, OpenAI, and Nvidia interviews.