HAI-HKUST（香港科技大学前沿人工智能社团） 是一个围绕前沿人工智能研究、经验共享与科研合作持续共建的学生学术社群。

我们希望连接对人工智能感兴趣的研究者、工程实践者与同学，通过高质量的学术分享、长期沉淀的公共知识以及轻量开放的科研协作，让一次讨论能够继续成为一份经验、一项实践，甚至一个新的合作项目。

分享带来视野，经验减少信息差，合作将想法转化为实际成果。

邀请校内外研究者与工程实践者，围绕前沿方向分享问题意识、方法选择、研究过程与真实经验。

我们希望每一次分享都不仅展示结果，也能够讨论：

• 为什么选择这个问题；
• 方法是如何形成的；
• 哪些尝试没有奏效；
• 当前工作的边界在哪里；
• 下一步还能继续探索什么。

→ 查看学术分享

持续沉淀申请、科研入门、项目实践、实习与发展路径中的真实经验。

相比只记录最终结果，我们更希望保留每个关键选择背后的信息、时间线、准备过程、项目积累、失败经验与反思，让个人经历逐渐成为后来者可以检索、比较和复用的公共资源。

→ 浏览飞跃手册

建立轻量开放的想法交流、论文讨论、复现与项目协作机制。

科研合作不必从一个宏大的项目开始。一个清晰的问题、一次论文复现，或者一个两周的小实验，都可能成为长期合作的起点。

我们鼓励从明确目标和可验证的小成果开始，在合作中逐渐建立信任。

→ 了解科研合作

我们的讨论与项目会围绕前沿人工智能持续扩展，目前重点关注：

Multimodal AI · Embodied Intelligence · World Models · AI for Science · Generative Models · Trustworthy AI

我们也欢迎人工智能与机器人、系统、科学、设计及其他领域之间的跨方向探索。

这里是 HAI-HKUST 的 GitHub 开源协作空间。

除了社团官方网站，我们希望逐步在这里沉淀：

• 技术项目与实验性探索；
• 论文复现与工程实践；
• 学术活动相关代码与资料；
• 可公开复用的学习与研究资源；
• 社团网站与技术基础设施；
• 由成员共同维护的开源项目。

GitHub 不只是代码的存放位置，也将成为社团长期共建和知识沉淀的一部分。

我们欢迎不同年级、专业和技术背景的同学参与。

你不需要已经成为某个方向的专家。一个清晰的问题、一篇认真准备的论文、一段真实的项目经验，或者一次小而明确的代码改进，都可以成为有价值的贡献。

在 GitHub 上，我们鼓励：

• 从小而明确的问题开始；
• 保持 Pull Request 范围清晰；
• 说明修改的动机与影响；
• 尊重已有设计和项目结构；
• 让代码、文档与讨论能够被后来者理解；
• 将有价值的过程持续沉淀下来。

Small steps. Clear contributions. Shared knowledge.

你可以：

参加分享
了解新的研究方向，与讲者和其他成员交流。

贡献经验
将申请、科研、项目或实践中的真实过程沉淀为公共知识。

发起合作
带着一个具体问题、论文或实验寻找同行者。

参与开源
浏览我们的 Repositories，通过 Issue 或 Pull Request 参与项目建设。

→ 访问官方网站
→ 浏览全部仓库

HAI-HKUST — HKUST(GZ) Frontier AI Club is a student-driven academic community built around frontier AI research, knowledge sharing, and research collaboration.

We connect students, researchers, and practitioners interested in artificial intelligence through research talks, shared knowledge, and lightweight collaboration.

Our goal is to help a conversation grow into reusable knowledge, a practical experiment, or even a long-term research collaboration.

Sharing broadens perspectives. Experience reduces information gaps. Collaboration turns ideas into outcomes.

We invite researchers and practitioners to share not only their results, but also the reasoning and process behind their work.

We care about questions such as:

• Why does the problem matter?
• How was the approach developed?
• What did not work?
• Where are the current limitations?
• What could be explored next?

→ Explore Research Talks

The AI Feiyue Handbook is a continuously evolving collection of real experiences in applications, research, projects, internships, and career development.

Rather than presenting outcomes alone, we encourage contributors to document timelines, decisions, preparation, failures, trade-offs, and reflections so that individual experience can become reusable public knowledge.

→ Browse the Handbook

We provide lightweight ways to connect around ideas, papers, reproduction tasks, and early-stage projects.

Research collaboration does not need to begin with a large project. A focused question, a paper reproduction, or a two-week experiment can become the starting point of a longer collaboration.

→ Explore Research Collaboration

Our current areas of interest include:

Multimodal AI · Embodied Intelligence · World Models · AI for Science · Generative Models · Trustworthy AI

We also welcome interdisciplinary exploration across AI, robotics, systems, science, design, and beyond.

This organization is the open-source collaboration space of HAI-HKUST.

Alongside our website, we aim to gradually build and maintain:

• technical and experimental projects;
• paper reproductions and engineering work;
• resources from academic activities;
• reusable learning and research materials;
• community infrastructure;
• member-driven open-source projects.

GitHub is not only where our code lives. It is also part of how we document, collaborate, and build shared knowledge over time.

Students from different disciplines, backgrounds, and levels of experience are welcome to contribute.

You do not need to be an expert first.

A clear question, a carefully studied paper, an honest project retrospective, or a small and focused code improvement can all become meaningful contributions.

We encourage contributions that are:

• focused in scope;
• clear in motivation;
• respectful of existing project structure;
• easy to review and understand;
• useful to future contributors.

Small steps. Clear contributions. Shared knowledge.

Join a research talk.
Contribute an experience.
Start a collaboration.
Build something open source.

Website · Repositories