汪蓄辉
Building education tools and AI research that make learning more intuitive.
I'm a high school senior at BASIS Independent Silicon Valley '27 and co-founder of Stellar Learning, where I build free learning tools, visual explanations, and student programs for more accessible education.
Organizations
Organizations focused on education products, research systems, and student-led programs for nonprofit impact.
Stellar Learning
Main platform for building and testing free AI-driven learning tools, from visual explanations and generated practice to curriculum-aligned courses.
BISV Hacks
Inclusive student-led hackathon at BASIS Independent Silicon Valley Upper School, designed to make building technology more approachable for students across the Bay Area.
Math For Real
Student volunteer organization helping students see math as intuitive, discussable, and connected to the real world.
Research
Many of these papers connect to systems built for Stellar Learning, including visual explanations, course generation, and reliable question generation.
Introduces GraphGPT, combining causal graphs with GPT-style prediction to generate logical steps and diagrams for intuitive education, with reported 25% higher multiple-choice accuracy than ChatGPT.
Explores arbitrary-size video generation using existing diffusion models, video outpainting, and frame interpolation to improve spatial and temporal flexibility without expensive retraining.
Presents Stellar Nova, a visually-first education assistant that adds interactive explanations and gamification to make complex concepts more intuitive and engaging.
Introduces CourseGen, a multi-agent framework behind Stellar Nova for helping course developers create intuitive online courses at scale.
Highlights interactive GenAI for intuitive learning through scientific graph generation and an AI-human course creation framework implemented in Stellar Nova.
Uses causal graph-guided chain-of-thought reasoning and a multi-agent architecture to generate accurate, curriculum-aligned questions with up to 70% quality improvement.
Frames educational question generation as a hallucination-minimization problem using staged agents, validation, and refinement, reducing hallucinations by over 90% in evaluation.
Explores lightweight preference alignment for automatic question generation using a small diagnostic set, supervised fine-tuning, and DPO, improving outputs by up to 13%.
Unifies causal graph-guided reasoning with multi-agent hallucination mitigation for trustworthy question-answer generation, showing 90% hallucination reduction and 70% quality improvement.
Profile
Academic, technical, and multilingual context behind the education technology work.
Technical Stack
Technologies used for learning platforms, research prototypes, and web products.
Recognition
Recognition across math, AP coursework, and piano performance.
Mathematics
Invitational round of the AMC series; AIME invites at least the top 2.5% of AMC 10 scorers and top 5% of AMC 12 scorers.Advanced Placement
Requires a 3.5+ average across all AP exams taken and scores of 3 or higher on at least five exams.Piano
First-place placement in an adjudicated international music competition, evaluated by instrument, age division, and recorded performance.Piano
Second-place placement in American Protégé’s international Romantic music competition, judged by age category and video performance.Academic Background
Advanced school coursework and external study across STEM, humanities, language, music, and curriculum design.
Additional coursework
Additional coursework
Additional coursework
Additional coursework
Communication
English, Mandarin, and Shanghainese are used for everyday communication, while Classical Latin comes through years of studying literature and texts.
English
Hello, I'm Nicholas.
Learning should be clear, intuitive, and open to everyone.
普通话 (中文)
你好，我是汪蓄辉。
学习应该清晰、直观，并让每个人都能获得。
上海闲话 (中文)
侬好，我是汪蓄辉。
读书要看得懂，便当，阿要大家侪学得到。
Lingua Latina
Salve, mihi nomen Nicolaus est.
Faciamus ut discere omnibus clarius faciliusque sit.
Mandarin and Shanghainese are two spoken varieties of Chinese: Mandarin is the standard variety associated with Beijing pronunciation, while Shanghainese is spoken in and around Shanghai.
