AI interview cram sheets built by agents while you sleep

What it does ARIS-in-AI-Offer is a curated set of 35 bilingual (Chinese/English) interview cheat sheets spanning ML, LLMs, diffusion, agents, and world models. Each sheet is auto-generated by the ARIS research-agent workflow and rendered into a single HTML file with live MathJax formulas and highlighted PyTorch code. The project is built for China’s 秋招 — the annual AI campus recruiting season — and also includes a CV-to-academic-homepage generator and hand-authored long-form surveys.

The interesting bit Every sheet follows the same rigid three-pillar structure: foundations with full formula derivations, 25 stratified interview questions (L1/L2/L3), and runnable from-scratch PyTorch code. Quality control is deliberately adversarial — cross-model review rounds (Claude, GPT-5.6-sol, Gemini) catch bugs in estimators and derivations before the HTML ever hits GitHub Pages.

Key highlights

• 35 first-party cheat sheets across 8 categories, from Transformer blocks to World Models and Quantization
• Single-file HTML output with no backend required: fully offline, responsive, searchable, and phone-friendly
• Bilingual by design, with a unified index at easyaioffer.github.io that includes dark mode and a per-reader reading tracker
• Includes ARIS-Homepage, which turns a CV into a DBLP-fact-checked academic homepage via the same render pipeline
• Production pedigree: built on the ARIS agent stack (the main repo claims ~10k stars and a HuggingFace Daily Papers #1 spot)

Verdict Bookmark it if you’re prepping for Chinese AI campus recruiting or want to see agent-generated technical pedagogy at production scale. Skip it if you need quick API docs or non-ML interview material.