GenAI & LLM interview guides.
Go beyond transformer definitions. Prepare for architecture, evaluation, serving, retrieval, agents, security, and reliability.
RAG System Design Interview GuideDesign retrieval-augmented generation systems that are measurable, secure, low-latency, and robust to bad retrieval.LLM Evaluation Interview GuideBuild evaluation systems for generative AI that combine task metrics, model graders, humans, and online signals.AI Agents & Tool Use Interview GuideDesign reliable agent systems with explicit tool contracts, bounded autonomy, state, observability, and recovery.MCP Interview Guide: Architecture, Security & 2026 ChangesUnderstand Model Context Protocol architecture and the 2026 stateless, routable, authorization-hardened protocol core.Prompt Injection & Agent Security Interview GuideReason about prompt injection as an authorization and data-flow problem, not just a prompt-writing problem.KV Cache Interview GuideExplain KV caching, memory/latency trade-offs, long-context pressure, offloading, quantization, and prefix reuse.LLM Inference Optimization Interview GuideOptimize LLM serving across throughput, time-to-first-token, inter-token latency, memory, and cost.Speculative Decoding Interview GuideExplain how speculative decoding reduces inter-token latency and when draft/verification overhead is worth it.Prefix Caching Interview GuideUse prefix caching to skip repeated prompt prefill while reasoning about privacy, cache keys, and hit rate.RAG vs Fine-Tuning Interview GuideChoose among prompting, RAG, fine-tuning, and combinations based on where the missing capability actually lives.Embeddings, Vector Search & Reranking Interview GuideDesign retrieval stacks from embedding choice and ANN indexes through hybrid retrieval and reranking.LLM Observability & Reliability Interview GuideMonitor generative AI systems across traces, quality, tools, retrieval, latency, cost, safety, and drift.SFT vs DPO vs RLHF vs GRPO Interview GuideCompare modern LLM post-training methods by supervision signal, optimization loop, data requirements, stability, and evaluation.Mixture of Experts (MoE) Interview GuideExplain sparse expert routing, capacity, load balancing, communication, and serving trade-offs in modern MoE models.Multimodal LLM Systems Interview GuideDesign systems that combine text, image, audio, or video with modality-specific preprocessing, evaluation, and serving.LLM Guardrails & Model Routing Interview GuideDesign layered guardrails and routing across quality, latency, cost, policy risk, and fallback behavior.