# AI Engineering Interview Questions: A Cheat Sheet for LLM and Agent Roles > AI Engineering Interview Questions is a Markdown repository that collects questions and answers for roles built around large language models: AI Engineer, LLM Engineer, Agentic AI Engineer, LLMOps Engineer, and related titles. It is organized into 14 topic areas and maintained by Amit Shekhar, founder of Outcome School. **amitshekhariitbhu/ai-engineering-interview-questions** — Your Cheat Sheet for AI Engineering Interview – Questions and Answers. - Repository: https://github.com/amitshekhariitbhu/ai-engineering-interview-questions - Website: https://outcomeschool.com/program/ai-and-machine-learning - Stars: 3,254 · Forks: 590 - Language: Markdown - License: Apache-2.0 - Published: 2026-09-09 · Updated: 2026-09-09 · Language: en - Canonical page: https://hysenlabs.com/projects/amitshekhariitbhu-ai-engineering-interview-questions ## What This Repository Is and Who It Is For AI Engineering Interview Questions is a plain Markdown cheat sheet of questions and answers designed for candidates preparing for interviews at companies building AI products. The README lists the target roles: AI Engineer, Gen AI Engineer, LLM Engineer, Agentic AI Engineer, AI Agent Engineer, Forward Deployed Engineer, AI Solutions Architect, AI Platform Engineer, Applied AI Engineer, MLOps Engineer, and LLMOps Engineer. All of these roles center on working with large language models and the systems built around them. This is a different interview domain from classical machine learning, where typical questions involve statistics, optimization algorithms, and classical model architectures. The README's Must Know section puts the scope plainly: LLM, RAG, MCP, agents, fine-tuning, and quantization are the baseline topics an interviewee must cover. The repository is maintained by Amit Shekhar, who also runs Outcome School, and the README notes it will keep receiving new questions and answers. The last push was on 2026-09-19. ## Fourteen Topic Areas and How the Repository Is Organized The README is structured as a single Markdown document organized into 14 sections, each covering a distinct area of AI engineering work. The sections are: LLM Fundamentals, Prompt Engineering, Retrieval-Augmented Generation, AI Agents and Agentic Systems, Fine-Tuning and Model Adaptation, Vector Databases and Embeddings, AI System Design, LLMOps and Production AI, Evaluation and Testing, AI Safety Ethics and Responsible AI, Multimodal AI, AI Infrastructure and Scalability, Coding and Practical Implementation, and Behavioral and Scenario-Based Questions. Within each section, questions are listed with answers linked to external resources: YouTube videos on the Outcome School channel, blog posts on outcomeschool.com, LinkedIn posts, and Substack articles. Some questions have inline answers in the README; others point to an external explanation. For example, the LLM Fundamentals section includes questions like 'What is tokenization in LLMs?', 'Explain BPE (Byte Pair Encoding)', 'What is self-attention, and how does it work in Transformers?', and 'Walk me through what happens, step by step, in one forward pass of a decoder-only Transformer.' Each links to a detailed writeup or video. This means the repository is an index and a navigation map rather than a standalone reference. The answers live partly in the README and partly off-site. ## How to Navigate and Use the Repository The repository contains a single README.md file and an assets directory. There is no code to install or run. To access the full content, either browse the README on GitHub or clone the repository: Open the table of contents in the README to identify which sections are most relevant to the role you are preparing for. For most AI engineering interviews, the LLM Fundamentals, Prompt Engineering, RAG, and Agents sections are the highest-priority starting points. The Must Know list identifies six core concepts: LLM, RAG, MCP, agents, fine-tuning, and quantization, along with a YouTube video that explains all six in one place. For questions where the answer is a link rather than inline text, follow the link. The external resources range from short LinkedIn posts explaining a single concept to full blog posts on outcomeschool.com covering topics like the mathematics behind attention mechanisms, causal masking, and positional encodings. The Coding and Practical Implementation section is distinct from the other sections in that it covers hands-on tasks rather than conceptual explanations, and the Behavioral and Scenario-Based section covers the non-technical portion of interviews. ## What the Repository Covers in Depth The LLM Fundamentals section is the most detailed part of the README. It covers transformer architecture from the ground up: the forward pass, tokenization methods (BPE, WordPiece, SentencePiece), attention mechanisms (self-attention, cross-attention, multi-head attention, causal masking), positional encodings, residual connections, feed-forward networks, context windows, temperature, top-k and top-p sampling, greedy decoding, and beam search. Most of these questions link to dedicated explanations. The RAG section addresses retrieval-augmented generation design questions, and the Vector Databases and Embeddings section covers the data layer that underpins RAG systems. The AI System Design section addresses architectural decisions, which appear frequently in senior-level interviews. The LLMOps and Production AI section covers deployment concerns: latency, cost, monitoring, and the operational differences between running a model in a notebook and running it in a production API. The AI Safety section covers alignment, responsible deployment, and ethics questions that appear at companies with active safety programs. The Multimodal AI and AI Infrastructure sections address newer interview areas that have expanded as more companies deploy multimodal models and build inference infrastructure. ## What Is Missing and Where to Set Expectations The repository does not contain inline code. None of the answers include Python implementations, configuration files, or worked examples that a candidate could run locally. For questions about practical coding tasks, the answers point to blog posts or videos rather than providing code directly in the README. Answers vary in depth. Some questions receive a link with no inline text at all; others receive a short explanation alongside the link. A candidate who needs a self-contained reference for offline study will find gaps. The resource functions better as a structured checklist to guide study than as a complete answer key. The repository also has no exercises, no flashcard system, and no way to test knowledge retention. It is a reading list organized by topic. Teams that run internal mock interviews or structured study groups will need to build the practice layer themselves. ## How This Compares to Classical ML Interview Prep Resources focused on classical machine learning and data science interviews, such as Chip Huyen's Machine Learning Interviews Book, cover statistics, linear algebra, optimization theory, classical model architectures, and ML system design for production systems built on traditional models. Those resources remain relevant for data scientist and ML engineer roles at companies that train their own models from scratch. This repository covers a different set of skills: working with pre-trained LLMs through APIs, designing retrieval systems, building agent pipelines, and operating LLM-based products in production. The overlap between the two interview domains is partial. LLM fundamentals and transformer internals appear in both; statistics, classical model training, and data pipeline engineering appear more in classical ML prep than here. An engineer applying for an LLM-focused role who has only studied classical ML prep may be underprepared for questions about attention patterns, RAG retrieval strategies, prompt engineering trade-offs, and agent evaluation. This repository directly targets that gap. ## Conclusion This repository is the right resource for engineers preparing specifically for LLM-era roles such as AI Engineer, Agentic AI Engineer, or LLMOps Engineer, where interviewers expect fluency in transformer internals, RAG design, agent architectures, and production concerns like evaluation and safety. It is a less complete resource for data science or classical ML roles where linear algebra, statistics, and traditional model training questions dominate. The repository does not have an install step or any executable code: access it on GitHub or clone it to browse locally. Before an interview, verify that the must-know list covering LLM, RAG, MCP, agents, fine-tuning, and quantization is familiar territory, since the README lists these as the baseline. ## FAQ ### How to prepare for an AI engineering interview? The repository recommends starting with its Must Know section covering LLM, RAG, MCP, agents, fine-tuning, and quantization. The README links to a YouTube video explaining all six. From there, work through the 14 topic sections, following the linked resources for each question. ### What are some common AI engineering interview questions? The repository covers questions on transformer architecture (attention, tokenization, positional encoding), RAG design, agent architectures, fine-tuning and model adaptation, vector databases, LLMOps, evaluation, and AI safety. The Must Know baseline is LLM, RAG, MCP, agents, fine-tuning, and quantization. ### What roles does this interview question repository target? The README lists AI Engineer, Gen AI Engineer, LLM Engineer, Agentic AI Engineer, AI Agent Engineer, Forward Deployed Engineer, AI Solutions Architect, AI Platform Engineer, Applied AI Engineer, MLOps Engineer, and LLMOps Engineer as the target roles. ## Sources - [amitshekhariitbhu/ai-engineering-interview-questions on GitHub](https://github.com/amitshekhariitbhu/ai-engineering-interview-questions) - [Issues](https://github.com/amitshekhariitbhu/ai-engineering-interview-questions/issues) - [License: Apache-2.0](https://github.com/amitshekhariitbhu/ai-engineering-interview-questions/blob/main/LICENSE) - [Project website](https://outcomeschool.com/program/ai-and-machine-learning) - [README](https://github.com/amitshekhariitbhu/ai-engineering-interview-questions/blob/main/README.md) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/amitshekhariitbhu-ai-engineering-interview-questions