AI, LLM & Machine Learning Interview Practice

Practice real technical interview questions across 12 core domains. Inspect in-depth answers, follow-up questions, key architectural concepts, and test live mechanics with interactive visualizers.

SHOWING 8 QUESTIONS0 Mastered

LLM & Generative AISenior

Explain how the Attention mechanism scales with sequence length N, and what optimizations mitigate this bottleneck.

#Quadratic complexity O(N²)#FlashAttention SRAM tiling#KV-cache bottleneck#Grouped-Query Attention (GQA)

RAG ArchitectureSenior

What are the main causes of retrieval failure in RAG systems, and how do you architect solutions for them?

#Hybrid Search (Dense + Sparse)#Parent-Document chunking#Cross-Encoder Reranking#Lost in the middle phenomenon

Fine-Tuning & LoRASenior

How does Low-Rank Adaptation (LoRA) work mathematically, and why does it not increase inference latency?

#Low-rank decomposition#Rank r hyperparameter#Weight merging#Memory reduction during backprop

ML Metrics & EvaluationMid

Explain the trade-off between Precision and Recall. In what scenarios would you prioritize one over the other?

#False Positives vs False Negatives#F1-Score harmonic mean#Threshold tuning#Confusion Matrix

Machine LearningMid

What is the Bias-Variance Tradeoff, and how do you diagnose whether a model is underfitting or overfitting?

#Generalization gap#High bias = underfitting#High variance = overfitting#Learning curves

AI System DesignPrincipal

Design a low-latency, scalable AI Agent architecture for an e-commerce support system handling 10,000 requests/minute.

#Semantic caching#Intent routing#PagedAttention / vLLM#Strict execution bounds#Guardrails

Embeddings & VectorsMid

What is the difference between Cosine Similarity, Dot Product, and Euclidean Distance for vector search?

#L2 normalization trick#Cosine angle vs magnitude#SIMD dot product performance#Nearest neighbor indexing

Prompt EngineeringJunior

What is Chain-of-Thought (CoT) prompting, and why does it significantly improve reasoning on multi-step math and logic problems?

#Autoregressive conditioning#Working scratchpad memory#Zero-shot CoT ("Let's think step by step")#Self-consistency