Rui Yang received a bachelor's degree in Electronic Information Engineering from CUHK-Shenzhen and an MA in Chinese Historical Studies from HKU. She is interested in modern Chinese history, the history of science and technology, digital humanities, and AI-assisted programming.
Tools and platforms she has built: ray-bot-ai.github.io, including a full-text search of the Shanghai Municipal Police Files, 1894–1949: ray-bot-ai.github.io/smp-files.
A re-transcribed, bilingual, full-text searchable edition of NARA microfilm publication M1750. Every page was re-read at native scanning resolution by a vision language model, then translated into Chinese; the original scan, the transcription and the translation sit side by side. Search covers English and Chinese together, with simplified／traditional／variant forms treated as equivalent. The edition documents its own error modes openly.
美国国家档案馆 M1750 缩微档案的重制转录本与中英对照全文检索库。逐页以原生分辨率 重新转录并译为中文，原件影像、转录、译文三栏并列；检索同时覆盖中英文，简繁异体等价。 本站公开自身的错误模式与质量边界。
A hand-verified index of free digital resources across China's national and provincial public libraries — whether you can register a library card online, what databases each library holds, and which of them are reachable remotely. Searchable, and browsable by resource, library, or category.
汇总国家图书馆及各省、直辖市、自治区公共图书馆：能否线上注册读者证、注册指南、 有哪些数字资源、各资源能否远程访问。支持检索与按资源／图书馆／分类浏览，数据经人工全量核对。
Full-text search across a vault that treats simplified, traditional, variant, and Japanese old／new character forms as equivalent — so Sinitic sources are no longer missed over differences in glyph form alone.
把简繁体、异体字、日本新旧字体视为等价，进行全库全文检索——处理汉文史料时不再因字形差异漏检。
A dual-engine OCR launcher — PaddleOCR-VL in the cloud and Apple Vision locally — for batch-transcribing PDFs and images into an Obsidian vault.
PaddleOCR-VL（云端）+ Apple Vision（本地）双引擎 OCR 调度，把 PDF／图片批量转录进 Obsidian 库。
Transcribes hard-to-read PDFs and images with a vision language model into companion Markdown notes — paginated against truncation, concurrent, with manageable prompt templates.
用视觉大语言模型转录难识别的 PDF／图片，生成伴生 md；分页防压缩、并发、可管理提示词模板。
Translates whole Markdown notes through a language model into companion translations — chunked against truncation, concurrent, multi-provider, with prompt templates.
整篇 md 调大语言模型翻译，生成伴生译文；分块防压缩、并发、可管理提示词模板，支持多供应商。
A Chinese／English／Japanese academic-literature search workflow for AI agents — query expansion around a research question, cross-database coverage, and tiered organisation of what comes back.
面向 AI 智能体的中英日多语种人文学术论文检索工作流——围绕选题扩展检索式、跨库补查、分级整理。