What AI Engineering Interviews Actually Ask, Company by Company
You've got an interview at a frontier AI lab next week. You know the job description lists things like "LLM inference" and "agentic systems," but what does that actually mean in the room? You can grind LeetCode all day, but that won't tell you whether Anthropic asks about transformer internals or whether Groq cares more about GPU serving. That's the gap this repository is trying to fill: real AI engineering interview questions, gathered from 35 companies and organized company by company.
What It Does
At its core, this is a cheat sheet for AI engineering interviews at top AI companies. The README describes it as a collection of real interview questions asked in AI engineering interviews at 35 companies, organized company by company, with answers linked wherever the maintainers have them. It's maintained by Outcome School, an AI and machine learning education program, and they've noted they'll keep updating it with new questions and answers.
The structure is where it earns its keep. Before you even get to the company-specific sections, there's a set of common questions asked across companies, broken into topic areas: LLM internals and architecture, inference and GPU performance, RAG and retrieval, agents and tool use, fine-tuning and alignment, evaluation and observability, safety and security, multimodal and voice AI, AI system design, and coding and data structures. Then the companies are grouped by category, which tells you something about how the industry itself is segmented.
The company list is broad. Frontier labs include Anthropic, OpenAI, Google DeepMind, Meta, xAI, Mistral, Cohere, DeepSeek, Moonshot AI, Zhipu AI, Alibaba (Qwen), and Sarvam AI. Big tech AI orgs cover Microsoft, Amazon, Apple, NVIDIA, Tesla, and a combined section for consumer-scale ML companies like Uber, Netflix, LinkedIn, Airbnb, Pinterest, and Spotify. There's a dedicated section for AI infrastructure players (Databricks, Groq, Together AI, Hugging Face, Scale AI, Perplexity), AI-native product companies (Cursor, Cognition, Sierra, Harvey, Glean, Character.AI, ElevenLabs, Abridge, Figure AI, Waymo), and forward-deployed/enterprise AI with Palantir. The README also spells out the range of roles these questions apply to, from AI Engineer and LLM Engineer through to Applied Scientist, Forward Deployed Engineer, and LLMOps Engineer.
Why It's Cool
-
It's organized the way you actually prepare. Most interview prep material is a flat list of questions. Here you can study by topic if you're strengthening weak areas, or by company if you've got a specific interview lined up. Those are two different modes of preparation, and this handles both.
-
The company grouping reveals industry structure. Splitting companies into frontier labs, big tech, infrastructure, AI-native products, and forward-deployed enterprise isn't just tidy bookkeeping. It hints at what each category values. A frontier lab interview probably leans hard on LLM internals and alignment; an infrastructure company like Groq or Together AI likely focuses on inference and GPU performance. The taxonomy itself is a useful signal.
-
The role list is refreshingly honest about the field. The README names fifteen role types, including some you don't see everywhere, like Forward Deployed Engineer and LLM Inference and Performance Engineer. If you're job hunting in AI right now, that list alone is a decent map of where the actual openings are.
-
Answers are linked where available. The README is upfront that this isn't a complete answer key. That honesty is better than pretending every question has a written solution. It tells you where you'll need to dig deeper yourself.
How to Try It
There's nothing to install here. It's a reference resource, so getting started is just reading.
- Head to the repository: github.com/pallavi-shekhar/ai-engineering-interview-questions-company-wise
- Start with the "Common Questions Asked Across Companies" section if you want a broad foundation across topics.
- Jump to the specific company section if you have an interview scheduled.
- For topic-wise questions and answers, the README points to a companion repo, AI Engineering Interview Questions and Answers, which is worth bookmarking alongside this one.
- Star or watch the repo if you want to catch updates, since the maintainers say they'll keep adding questions and answers.
Final Thoughts
This is a practical, no-frills resource, and that's the point. It won't teach you the material from scratch (you'll still need to actually understand how attention works or how KV caching affects throughput), but it will tell you what to study and where the bar sits at specific companies. It's best suited for engineers already working in or adjacent to AI who are preparing for interviews, especially those targeting multiple companies at once. The linked-answers caveat means some questions will send you searching elsewhere, but as a structured starting point for interview prep, it's a sensible one to keep open in a tab.
Follow @githubprojects for more developer tools and open source projects.