Free AI system design study guides drop
A new free GitHub repo walks senior candidates through AI system design topics—RAG, agents, inference, evals, MLOps and security—while separate repos offer 300+ curated LLM Q&As and a full AI curriculum roadmap mapping prompts to transformers. These resources bundle interview prep and production patterns into structured, shareable learning paths. (x.com 1) (x.com 2) (x.com 3)
The GitHub repository girijesh-ai/ai-interview-codex hosts a "LLM‑ML‑SYSTEM‑DESIGN‑MASTER‑GUIDE" plus a MASTER‑STUDY‑SCHEDULE that organizes system‑design examples, an LLM production guide, MLOps playbooks and an interview prep index. (github.com) The LLM production guide in that repo calls out specific inference optimizations—vLLM vs TGI (noted as a 23× throughput improvement), PagedAttention (≈80% KV‑cache reduction), FlashAttention (2–4× speedups), and quantization formats AWQ/GPTQ/GGUF alongside speculative decoding techniques. (github.com) Security and reliability material is explicit: the codex includes prompt‑injection detection, jailbreaking defenses, multi‑layer content‑moderation patterns and a dedicated RAG evaluation guide. (github.com) Interview Q&A scale is split across projects: a browsable "LLM Interview Questions" guide adapted from Hao Hoang compiles 315 items in a study browser, useful for systematic practice. (shanselman.github.io) Other community collections include llmgenai/LLMInterviewQuestions (100+ curated questions and ~1.7k GitHub stars) and KalyanKS‑NLP's LLM Interview Q&A Hub (100+ Q&As, linked RAG and prompt‑engineering resources, ~831 stars). (github.com) The rohitg00 "AI Engineering from Scratch" curriculum maps a 20‑phase roadmap with 230+ lessons to an outputs/ directory that stores prompt templates and executable "skills" designed to translate prompts into transformer‑based implementations. (deepwiki.com) The master guide has visible forks for reuse—ivesh/ai‑interview is a public fork of girijesh‑ai's codex—illustrating community reuse and shareable study paths across contributors. (github.com)
