AIMLInterviews: an interview-prep guide with an MCP tutor for AI/ML roles
This repo is meant to serve as a guide for Machine Learning/AI technical interviews.
At a glance
- What is it?
- AIMLInterviews is a Markdown curriculum for AI/ML technical interviews at large companies, plus an MCP server that turns a compatible assistant into a coach. The content is opinionated and the MCP path is the newest, least documented part.
- Who is it for?
- AIMLInterviews fits engineers and applied scientists targeting AI/ML roles at large companies who want a chaptered reading list and are willing to clone the repository to use the MCP tutor. It is a poor fit if you want a runnable codebase, a question bank with graded answers, or a data science or research scientist track, since the README says those roles have different interview structures.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 6 days ago.
- What is it written in?
- Mainly Jupyter Notebook, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 28, 2026, and from our analysis. They are not legal advice.
Editorial analysis
What AIMLInterviews covers and who it is written for
The repository is a guide, not a library. Its README states the goal plainly: it "aims to serve as a guide to prepare for AI and ML Technical interviews" for roles at big tech companies, with FAANG named specifically. The author's own history is the stated basis for the content, drawn from preparation that produced offers from Meta, Google, Amazon, Apple and Roku in 2020 and repeat offers from Amazon and Apple in 2025.
The material is split into numbered chapters, each a Markdown file under src/: general coding and data structures, ML coding, ML fundamentals, and ML/GenAI/LLM system design. Agentic AI systems live in a separate repository. A behavioral chapter ships with two extra artifacts, a Google Sheets worksheet and an Excel template in src/behavioral/. A resources chapter points at GenAI learning material.
The audience is narrower than the title suggests. The README says the guide focuses on AI/ML engineering, applied science and tech lead roles at big companies, and that adjacent roles such as data science or research scientist have different interview structures, even if some modules transfer. Startup interviews get one sentence: they are usually tailored to the company's own problems rather than following the big-company pattern.
How the repository is organized and what the MCP server adds
Reading the repository is the default path. The README presents a table mapping chapters to files, so the reader moves from coding drills to ML fundamentals to system design. Everything is Markdown and Jupyter Notebook files, which means the content is readable on GitHub without cloning and renders as plain text in any editor.
The second path is programmatic. The README describes an AI tutor delivered as an MCP server named aimlinterviews-mcp, published so it can be run with npx. According to the README, it "discovers curriculum problems, offers progressive hints, builds learning and company preparation plans, and reviews answers using a no-spoilers teaching style." Those four capabilities are the tool surface. The server reads the curriculum from a local clone, which is why the README tells you to clone the repository first and either run the command from inside the clone or set AIMLINTERVIEWS_ROOT. The README points to MCP/README.md for configuration, tools, and development instructions.
That is the whole architecture as documented at the top level: a static content tree plus a server that indexes it and exposes tools to an assistant. The README defers configuration, the tool list and development instructions to MCP/README.md, so the top-level file is a pointer rather than a specification.
Installing the MCP tutor and running a first session
The README gives two registration commands, one per assistant. Both assume the MCP client is already installed and that you have cloned the repository, because the server reads curriculum files from disk. Run the command for your client:
claude mcp add aimlinterviews -- npx -y aimlinterviews-mcpIf you use Codex instead, the README shows the equivalent registration under a different subcommand:
codex mcp add aimlinterviews -- npx -y aimlinterviews-mcpAfter registering, the client should list a server named aimlinterviews. The README says to clone the repository first, then run the command from the clone or set AIMLINTERVIEWS_ROOT, and it points at MCP/README.md for configuration, tools, and development instructions. With the server registered, the first useful session is a hint request rather than a full answer, since the stated teaching style withholds spoilers. Ask the assistant to pick a curriculum problem and give you a progressive hint. What you should see is the assistant calling the server's discovery and hint tools rather than answering from its own memory. If the tool calls do not appear, the server is not connected, and MCP/README.md is the place to check configuration.
The guide is prose, and that limits what it can verify
Nothing in the repository grades you. The chapters are explanations and notes, so a reader who works through them gets exposure to topics, not a signal about readiness. The only feedback loop the README describes is the MCP server's answer review, and that review runs inside whatever assistant you connected, using the server's no-spoilers style as a constraint rather than a rubric.
The behavioral chapter is the exception in form, not in function. It ships a worksheet and an Excel template, which implies self-assessment rather than automated scoring. The repository layout shows no test suite and no scoring script, and the README does not claim any.
There is also a commercial boundary worth stating. The top of the README advertises 1:1 coaching and mock interviews through aimlinterviews.io, and the repository is the free companion to that service. The chapters point at the paid offering for the part a static guide cannot supply, which is a human judging your answers under time pressure.
What the MCP server does not do, and when to skip it
The MCP path has a dependency the reading path does not: a local clone. The README instructs you to clone the repository first and either run from the clone or set AIMLINTERVIEWS_ROOT. That means the tutor cannot be used from a bare npx invocation on a machine without the content, and the top-level README does not document what happens when the root is wrong or missing. Rollback of a registration is not covered either, so removing the server means consulting your client's own documentation.
Version pinning is unresolved in the README. The command uses npx -y aimlinterviews-mcp with no version suffix, and no releases are listed for the repository, so there is nothing in the README to pin against. For a tool that reads a moving curriculum, that is a real trade-off: you always get the current server, and you cannot reproduce a past session.
Skip the MCP server if you only want to read. The chapters are plain Markdown and render fine on GitHub, and the extra setup buys you nothing until you want hints or a generated prep plan. Skip the whole repository if you are preparing for a data science or research scientist loop, which the README explicitly places outside its scope.
How it differs from the standard interview-prep alternatives
The closest comparison is a general interview-preparation book, such as the machine learning interviews book readers often search for. A book is edited, sequenced and stable, and it does not change between printings. AIMLInterviews is the opposite: the README says it was updated for 2026 with expanded LLM, multimodal AI, post-training and GenAI system-design content, and the last push to the repository was on 2026-09-02. That recency is the point, since LLM and agentic interview topics move faster than a print cycle.
The second alternative is a question bank with answers, the format behind most "interview questions and answers" searches. A bank optimizes for recall of specific answers. This repository optimizes for structure: chapters that mirror the modules the author observed across FAANG loops, plus an explicit statement that companies do not follow one unique structure. The MCP server pushes further in that direction with progressive hints instead of answers.
The third alternative is a dedicated agentic-AI resource. The README links Agentic AI Systems as a separate repository rather than folding it in, so the chaptered guide here stops at ML and GenAI system design and hands agentic material off.
Licence, maintenance and the cost of keeping up
The repository is MIT licensed, and the LICENSE file sits at the top level alongside the READMEs. MIT is permissive: you can reuse, modify and redistribute the content, including commercially, provided the copyright notice and permission notice are retained. That matters if you want to adapt chapters into internal onboarding material. It does not extend to the paid coaching service, which is a separate offering at aimlinterviews.io, and it does not cover the linked Google Sheets worksheet, which lives on a third-party service. This is a description of the licence text, not legal advice.
Maintenance is active by the only measure available: the repository is not archived, and the last push was on 2026-09-02. No releases are listed, so there is no versioned artifact to track and no changelog to read. The upgrade cost is therefore near zero for the reading path, since a git pull brings new chapters, and undefined for the MCP path, since the server is fetched fresh by npx each time. The real cost is editorial: you have to re-read chapters after a pull to know what changed, because there is no release note to tell you.
Translations add a second maintenance surface. The README links Simplified Chinese and Persian versions, README-CN.md and README-FA.md, with cn/ and fa/ directories alongside src/. The README does not state how those are kept in sync with the English content, so a reader relying on a translation should check the English file for the same chapter before trusting it.
Editorial conclusion
AIMLInterviews fits engineers and applied scientists targeting AI/ML roles at large companies who want a chaptered reading list and are willing to clone the repository to use the MCP tutor. It is a poor fit if you want a runnable codebase, a question bank with graded answers, or a data science or research scientist track, since the README says those roles have different interview structures. Before relying on it, check that README-CN.md and README-FA.md still match the English README, and confirm the MCP server's tool list in MCP/README.md, because the top-level README only shows the install command and defers everything else to that file.
Frequently asked questions
What is asked in an AI/ML interview?
The repository organizes its material around the modules it says are most common for technical ML roles: general coding and data structures, ML coding, ML fundamentals covering classic ML, LLMs and multimodal AI, and ML/GenAI/LLM system design. Agentic AI systems are covered in a separate linked repository, and behavioral interviews have their own chapter with a worksheet and Excel template.
Is an AI interview a red flag?
The repository does not address this question. Its README only describes the structure of AI/ML technical interview modules and the author's own preparation and offers, so there is no material here on judging whether a particular interview process is a warning sign.
Is AI ML hard to learn?
The README does not make a claim about difficulty. It says the guide focuses on AI/ML engineering, applied science and tech lead roles at big companies, and that interviewing is a skill whose results improve with practice.
What does an AI interview consist of?
The README states that AI and ML interviews at different companies do not follow a unique structure, but that the components it observed were very similar across FAANG companies. It lists coding, ML coding, ML fundamentals, system design and behavioral modules as the recurring parts, and notes that startup interviews are usually tailored to the company's own use cases.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/alirezadir-aimlinterviews)