What 1,704 Interview Questions Show About Big Tech Loops

We counted Devana's question bank: 1,704 interview questions across 46 companies. By round, 33% are coding, 21% system design, 28% technical deep dives and 18% behavioral. 46% are rated hard and none easy, and Scalability is the most common topic, on more than a third of them. We wrote these questions to reflect how each company interviews, so they show where preparation can focus, not what any company actually asked.

What we counted

Every question in Devana's company sets, as of November 1, 2026. Each one was written for Devana, from what that company and its candidates have said publicly about its products, values and interviews. None is a leaked or confirmed interview question. There is no session or user data here, and nothing about how candidates scored or whether they were hired.

The split by round type

Read the share column with care. 35 of the 46 sets have exactly 33 questions, and 24 of those split exactly 10 coding, 7 system design, 10 technical and 6 behavioral, because we built them to one shape. So it describes how the bank was built, not how often a company runs each round.

The hard column says more. System design is rated hard most often, at 67%, against 22% for behavioral. That fits how the round works: there is no single answer to reach, and the difficulty is in the follow-ups. See how a system design round is scored.

Questions per company

The other 35 sets have 33 questions each: Adobe, Atlassian, Block, Bloomberg, Cloudflare, Coinbase, Confluent, Databricks, Datadog, Discord, DoorDash, Dropbox, Elastic, Figma, GitHub, HashiCorp, Instacart, Lyft, MongoDB, Notion, Nvidia, Oracle, Palantir, Pinterest, Reddit, Robinhood, Salesforce, Shopify, Slack, Snap, Snowflake, Spotify, Tesla, Twilio, Vercel. A larger set means we wrote more for that company, not that it asks more questions.

The Uber, Airbnb and LinkedIn sets now have written guides too: the Uber interview process, the Airbnb interview process and the LinkedIn interview process. The company names in the table link to each mock interview page.

The most common topics

Look at what coding questions are tagged with. Scalability is on 199 of the 564, Databases on 164 and Concurrency on 158, ahead of Hash Map on 93, Strings on 90 and Graphs on 75. That is because we frame coding problems in the company's own domain, so the underlying skill is often standard but the question is about its systems. Being ready to say what happens at scale is part of answering one here.

What each set leans toward

Unlike the round split, this does vary, because it follows each company's products. These are the 12 sets that lean furthest from the bank as a whole:

That ranking compares each topic with how common it is everywhere. By share alone, the strongest are Oracle and MongoDB, with Databases on 82% and 79% of their sets. The larger sets lean less sharply: Stripe toward APIs (31% of its set), OpenAI toward APIs (32% of its set), Apple toward Frontend (22% of its set), Meta toward Hash Map (21% of its set) and Uber toward Graphs (21% of its set).

Which sets are rated hardest

Every question comes with follow-ups

All 1,704 questions carry follow-ups for the interviewer, 3.9 on average. We wrote them that way because the first answer is rarely the end of it, so practice being questioned about your answer as well as giving it. What to say when you don't know the answer covers the moment a follow-up lands somewhere you have not prepared.

Most questions, 1,220 of them, are for any software engineer rather than one specialty. The rest are written for a role: backend 231, data 53, security 51, DevOps 49, frontend 45, machine learning 32, mobile 21.

What these numbers cannot tell you

• What any company actually asks. We wrote these questions. For a company's own description of its process, read our guides, such as the OpenAI interview process, which cite it.
• How often a company runs each round. Most sets were built to the same split.
• How hard an interview is. The ratings and topic tags are our own judgment, applied by the people who wrote the questions.
• Whether any of this works. There is no session data here, and we make no claim about offers or pass rates.

How to use this

• Start from the lean. Find your target company above, and make sure you can talk through that topic out loud. Each company mock interview page lists its set.
• Practice scale inside coding. When you finish a problem, say what breaks at a hundred times the load. A coding mock interview gives you someone to say it to.
• Read the loop separately. For the stages and rounds themselves, see what big tech interview loops test and the interview guide.

Frequently asked questions