How many Mid Level questions are in this list?

100 questions, the highest rated Mid Level AI interview questions.

Are these real AI interview questions?

Yes. Every question is sourced from real AI interview reports across companies like OpenAI, Anthropic, Google, Microsoft, and Cohere, then reviewed by senior AI engineers for accuracy.

Do these questions come with detailed answers?

Yes. Every entry links to a full answer with explanations, hints, key tradeoffs, and recommended follow up questions. Most also include difficulty ratings and community quality scores.

How is the ranking calculated?

Questions are sorted by community quality score, with rating count as the tiebreaker, then difficulty (easier first). Quality scores come from interview candidates rating each question after working through it.

Who should use the Mid Level question list?

This list is curated for Mid Level AI interview candidates. Questions are selected based on community quality scores and cover the most common LLM, transformer, RAG, and production AI topics asked at that level.

Curated · Mid Level

Top 100 Mid Level Interview Questions

Here are the top 100 AI interview questions for Mid Level candidates, sorted by community quality. Each entry links to a detailed answer with explanations, hints, and source grade follow ups. Browse all Mid Level questions →

Sorted by community quality · refreshed as new questions are published.

Clear filter

Resize10 15 20 25 50 100

01
Explain scaled dot product attention.
Short AnswerMedium Decagon Freshworks
02
What is the KV cache in transformer inference?
FlashcardEasy Elevenlabs Phonepe
03
What is RLHF, and why is it used after pretraining?
Multiple ChoiceEasy Cognizant Flowise
04
What does LoRA do, and why is it popular for fine-tuning?
Multiple ChoiceMedium Anduril Rephrase Ai
05
In LLM serving, what is the primary driver of end to end latency for a generation request?
Multiple ChoiceMedium Meta NVIDIA
06
Which metric best measures whether a RAG answer is grounded in the retrieved context?
Multiple ChoiceMedium Datadog IBM
07
When should you fine-tune instead of using RAG?
Short AnswerMedium Kore Ai Persistent
08
Explain why transformers replaced RNNs for language modeling.
Short AnswerEasy Ironclad Together Ai
09
What does RAG primarily help with in LLM based applications?
Multiple ChoiceEasy Airbnb Perplexity
10
Why combine BM25 with dense embeddings for retrieval?
Multiple ChoiceEasy Anduril Crewai
11
PPO vs DPO, what's the practical difference?
FlashcardMedium Kpmg Sambanova
12
What is the Model Context Protocol (MCP) and what problem does it solve?
Multiple ChoiceEasy Accenture Dust
13
What distinguishes an AI agent from a plain LLM call?
Multiple ChoiceEasy Inflection Ai Mu Sigma
14
Which components make up a standard transformer block?
Multiple ChoiceEasy Goldman Sachs Mistral AI
15
Which of these are valid concerns when using LLM-as-judge for evaluation?
Multi-selectMedium Airbnb Labelbox
16
PremiumHNSW vs IVF, when…
FlashcardMedium Dust Goldman Sachs
17
How does MCP differ from OpenAI style function calling?
FlashcardMedium Phonepe Promptlayer
18
Match each vector index type to its primary characteristic.
Match PairsMedium Forethought Intel
19
Describe the typical agent execution loop.
Short AnswerMedium Forethought Runway
20
Which of these are valid considerations when choosing a chunking strategy for RAG?
Multi-selectMedium Ai21 Coinbase
21
Estimate the per call cost of a typical RAG chatbot using GPT-4o-mini.
FlashcardEasy OpenAI Pwc
22
Why is cosine similarity preferred over Euclidean distance for text embeddings?
FlashcardEasy Anthropic Cohere
23
Why does BPE tokenization use subwords instead of words or characters?
FlashcardEasy Meta OpenAI
24
What is the primary purpose of embeddings in modern NLP?
Multiple ChoiceEasy Amazon Perplexity
25
Spot the error in this explanation of temperature.
Spot the ErrorEasy Baseten Siemens
26
Put the steps of a basic RAG query pipeline in order.
Order StepsEasy Perplexity Pinecone
27
What is semantic chunking and when does it beat fixed size chunking?
FlashcardMedium Polyai Wipro
28
Predict the output: how many tokens does this string become with a typical BPE tokenizer?
Predict OutputEasy Fiddler Ai Pinterest
29
Fill in the scaled dot product attention formula.
Fill in BlankEasy Bytedance Gnani
30
What causes context pollution in a long running agent?
FlashcardEasy Amd IBM
31
Describe the three tier memory hierarchy used by modern LLM agent systems
FlashcardEasy Infosys Mu Sigma
32
Pick the curve shape that the lost-in-the-middle effect produces
Multiple ChoiceEasy Elastic Harvey
33
Why run agent generated code inside an isolated sandbox?
FlashcardEasy Midjourney Persistent
34
Where does an agent keep short term state during a single task?
FlashcardEasy Lepton Ai Salesforce
35
Why is context stuffing the opposite of context engineering, even on a 1M-token window?
FlashcardEasy Jane Street Moveworks
36
What does one agent trace capture step by step?
FlashcardEasy Polyai Sierra
37
PremiumPick the most distinctive…
Multiple ChoiceEasy Netflix Pinecone
38
Why cap each agent run with a cost per task budget?
FlashcardEasy Fractal Analytics Linkedin
39
How is long term memory typically implemented in an agent?
FlashcardEasy Fiddler Ai Roblox
40
Define the lost-in-the-middle effect and the empirical pattern it describes
FlashcardEasy Induced Ai Oracle
41
How does a sliding window memory keep a chat under budget, and what does it always lose?
FlashcardEasy Alibaba Infosys
42
Which OpenTelemetry GenAI attribute records the provider that served a call?
FlashcardEasy Krutrim Reliance Jio
43
What does observability mean for production LLM agents, and how is it different from app monitoring?
FlashcardEasy Tesla Two Sigma
44
Define trace, span, and parent span id in one tight pass
FlashcardEasy Bain IBM
45
When should a production agent pause for human approval?
FlashcardEasy Bain Cresta
46
Why do six short, specific instructions usually beat one long paragraph of guidance?
FlashcardEasy Anduril Cursor
47
Distinguish context engineering from prompt engineering in one practitioner's working definition
FlashcardEasy Ltimindtree Notion
48
Why is a 4,000-token system prompt usually a code smell?
FlashcardEasy Elevenlabs Flowise
49
Identify what Llama Guard 4 adds over Llama Guard 3
Multiple ChoiceEasy Redis Reliance Jio
50
Flashcard: what is a stop sequence in an LLM API call and what is it used for?
FlashcardEasy Deepseek Lyzr
51
Why do DeepSeek style stacks slice the FFN into many small experts instead of a few large ones?
FlashcardEasy Citadel Descript
52
Why run a toxicity classifier on output even when the model was RLHF tuned to refuse harmful content
FlashcardEasy Graphcore Neo4j
53
In MCP, what exactly is a 'tool' and who decides when to call it?
FlashcardEasy Midjourney Moveworks
54
Automated eval vs human eval: which is better for catching subtle factual errors in medical text?
Multiple ChoiceEasy Phonepe Unity
55
Watermarking and content provenance for AI generated images
FlashcardEasy Anduril C3 Ai
56
PremiumPick the biggest token…
Multiple ChoiceEasy Contextual Ai Dust
57
RoPE, spell out the acronym and name exactly which vectors it rotates
FlashcardEasy Gnani Mphasis
58
Order the stages of an invoice extraction pipeline from upload to approved record
Order StepsEasy IBM Siemens
59
Does FlashAttention change the attention output relative to standard attention?
Multiple ChoiceEasy Anduril Bain
60
Fill in the typical canary traffic percentages and soak duration for a prompt rollout
Fill in BlankEasy Crewai Mongodb
61
What does the dot product q . k actually measure inside an attention layer?
FlashcardEasy Lepton Ai Linkedin
62
What problem does a Q-Former or perceiver resampler solve inside a VLM connector?
FlashcardEasy Runway Tesla
63
Fill the blank: for a single head, QK^T has shape (T_query, ___).
Fill in BlankEasy Ai4bharat Autodesk
64
Why Hugging Face Hub functions as the de facto model registry for open weights
FlashcardEasy Autodesk Lepton Ai
65
A local stdio MCP server has no auth token. Is that a security hole?
FlashcardEasy Cresta Figure Ai
66
AlpacaEval 2 reports a 'win rate' for each model. Against what baseline and how is the winner decided?
Multiple ChoiceEasy Airbnb Bytedance
67
Flashcard: what is an agent 'trajectory' and why does it matter for evaluation?
FlashcardEasy Fractal Analytics Niki Ai
68
BLEU score comes up in a meeting. What does it measure and when does it still make sense?
FlashcardEasy Cognizant Comet Ml
69
The filesystem MCP server gives agents file access. How does it stop them from reading everything on disk?
FlashcardEasy Deepseek Dify
70
You have no reference answers for your eval set. Does that mean you cannot evaluate the model?
Multiple ChoiceEasy Replicate Sap
71
Select every artifact a prompt PR should produce in the CI run
Multi-selectEasy Jasper Lightning Ai
72
Why does Chatbot Arena (LMSYS) give a different picture of model quality than static benchmarks like MMLU?
Multiple ChoiceEasy Dify Notion
73
Describe the dual write pattern teams use during an embedding model swap.
FlashcardEasy Citadel Meesho
74
Order the stages of a hybrid sparse plus dense query with reranking inside the database layer.
Order StepsEasy Ai21 Infosys
75
Pick the items that are LLM application frameworks (not vector DBs, observability tools, or providers)
Multi-selectEasy Cresta Krutrim
76
What is LLM-as-judge and why did it emerge as an evaluation approach?
FlashcardEasy Adobe Airbnb
77
HELM from Stanford claims to be a 'holistic' evaluation. What makes it different from running MMLU alone?
Multiple ChoiceEasy Elastic Mphasis
78
Describe how end user thumbs up/down should flow back onto a trace
FlashcardEasy Citadel Mongodb
79
Order the lifecycle of a single extended thinking API call
Order StepsEasy Banana Dev Fractal Analytics
80
When does adopting LangChain pay for itself over a raw OpenAI SDK call?
FlashcardEasy Bcg Mu Sigma
81
GQA, decode the acronym and describe its KV sharing pattern
FlashcardEasy Accenture Tesla
82
Describe what happens during the decode phase of LLM inference.
FlashcardEasy Anduril Datarobot
83
Expand MQA and state its KV-cache cost tradeoff
FlashcardEasy Amd Meesho
84
Identify the B200 and GB200 NVL72 in NVIDIA's Blackwell lineup.
FlashcardEasy Browserbase LangChain
85
Walk through what actually happens during the prefill phase of an LLM forward pass.
FlashcardEasy Canva Induced Ai
86
Pick the right choice between rolling summary and vector store memory for cross-session continuity
Multiple ChoiceEasy Forethought Haptik
87
Choose the task where a verifiable reward can replace a learned reward model
Multiple ChoiceEasy Baidu Descript
88
What does 'persona prompting' (or role based prompting) actually do and where does it stop being useful?
Multiple ChoiceEasy Alibaba Flipkart
89
You hear about 'eval harnesses' like Promptfoo and DeepEval. Explain what an eval harness does that a Jupyter notebook cannot.
FlashcardEasy Intel Neptune Ai
90
Your colleague says MCP and function calling do the same thing. Where are they wrong?
Multiple ChoiceEasy Flipkart Notion
91
Flashcard: what does it mean when an LLM 'makes a tool call'?
FlashcardEasy Elastic Kpmg
92
Prompt caching shows up on Anthropic and OpenAI pricing pages, what does it actually do?
FlashcardEasy Palantir Robinhood
93
Why log the assembled context, not just the model response, in production?
FlashcardEasy Cred Oracle
94
Find the flaw: lower pretraining perplexity guarantees better chat quality
Spot the ErrorEasy Anthropic OpenAI
95
Describe the W_O projection in multi-head attention, its shape and what it mixes.
FlashcardEasy Cognizant LangChain
96
Why does a cross-encoder reranker pay off even though it is slower than a bi-encoder retriever?
FlashcardEasy Mckinsey Pinecone
97
Flashcard: why break a complex goal into subtasks before acting?
FlashcardEasy Character Ai Hcl
98
What is AutoGen and how does it coordinate multiple agents?
FlashcardEasy Phonepe Snap
99
PremiumMisconception: 'LoRA is just…
Multiple ChoiceEasy Contextual Ai Vellum
100
Replies tripled in length after RLHF with flat quality: pick the likely cause
Multiple ChoiceEasy Amd Glean

Related top picks

Change experience level

Fresher Senior Career Switcher

By topic

Top 100 Fine-Tuning Interview Questions→Top 100 Inference Optimization Interview Questions→Top 100 LLM Evaluation Interview Questions→

By role

Top 100 GenAI Engineer Interview Questions→Top 100 RAG Engineer Interview Questions→Top 100 AI Product Engineer Interview Questions→