张涵哲
Hanzhe Zhang
游戏设计师 | 游戏设计硕士 ’26
Game Designer | MS Game Design ’26
技术策划 AI + 游戏 系统设计 👋 联系我
Technical Design AI + Games Systems Design 👋 Contact
↓ 下滑查看详情
↓ Scroll for details
Projects Detail
MagiScape Screenshot 1
MagiScape Screenshot 2
麦琪的花园 LLM驱动的生存建造游戏
MagiScape LLM-Driven Survival Builder
GenAI系统策划 & 任务策划
GenAI Systems Design & Quest Design
一款俯视角2D生存建造游戏。玩家通过输入提示词和参考形象招募NPC,系统据此生成语音、立绘、人格和技能。每个NPC拥有多层记忆的AI智能体,能基于玩家行为和世界状态生成专属任务线。
A top-down 2D survival builder. Players recruit NPCs by inputting prompts and reference images, and the system generates their voice, portrait, personality, and skills. Each NPC is an AI agent with multi-layered memory, capable of perceiving, planning, and reflecting, generating exclusive personal quest lines based on player actions and world states.
AI任务生成系统重构
重新设计任务生成架构,构建结构化上下文环境,让AI感知玩家进度、NPC记忆和世界状态,在约束空间内按需生成。重构后内容量与稳定性显著提升。
LLM管线与提示词架构
设计角色生成与任务生成的完整LLM管线。通过分层上下文注入和按需查询的工具链约束生成边界。
任务内容设计与叙事
设计NPC个人任务线的叙事走向与解锁节奏,让任务承载NPC性格的同时与玩家主线产生呼应。另主导宣传PV脚本与录制。
AI Quest Generation System Redesign
Redesigned the quest generation architecture by building a structured context environment. This allows the AI to perceive player progress, NPC memory, and world states, generating quests on-demand within constraints. Content volume and stability significantly improved post-refactor.
LLM Pipeline & Prompt Architecture
Designed the complete LLM pipeline for character and quest generation. Constrained generation boundaries through layered context injection and an on-demand query toolchain.
Quest Content Design & Narrative
Designed the narrative direction and unlock pacing for NPC personal quest lines, ensuring quests carry the NPC's personality while echoing the player's main progression. Additionally, led the scripting and recording of the promotional PV.
Widget Idle Screenshot 1
Widget Idle Screenshot 2
Widget Idle Screenshot 3
Widget Idle Screenshot 4
Widget Idle 小组件主题桌面放置游戏
Widget Idle Widget-Themed Desktop Idle Game
系统设计 & 全栈开发
Systems Design & Full-Stack Dev
一款悬浮在桌面一角的放置游戏,将玩家使用电脑的自然行为与放置玩法结合。解锁键盘计数器,每次敲击都出产“构件”;开启番茄钟,专注工作的同时加速产出;接受喝水提醒,喝水来换取即时收益。实用与玩法,在每个小组件中自然合一。
An idle game that floats in the corner of your desktop, pairing natural computer use with idle mechanics. Unlock a keystroke counter—every keypress yields “components”; start a Pomodoro to boost output while you focus on work; accept hydration reminders and drink water for instant rewards. Utility and gameplay come together naturally in each widget.
核心体验设计
以"不占用注意力的有意义成长"为核心体验目标,面向一边工作一边挂机的桌面用户,围绕这一目标设计各类小组件的功能与交互——让玩家在正常使用电脑的过程中,自然地推进游戏进度。
经济系统与数值设计
围绕核心体验搭建完整的数值框架,涵盖点击收益、组件产出与升级成本曲线。通过解锁节奏和升级时机的控制,引导玩家从主动点击自然过渡到挂机生产。
技术实现
基于 React + TypeScript + Electron 架构,借助 LLM 辅助编程,主导完成全栈开发。
Core Experience Design
Aimed for “meaningful growth without stealing attention” as the core experience, for desktop players who idle alongside work. Shaped each widget’s features and interactions around that goal—so progress advances naturally as people use their computer normally.
Economic System & Numerical Design
Built a full numerical framework around that experience, covering click yields, widget production, and upgrade cost curves. Unlock pacing and upgrade timing guide the shift from active clicking to idle production.
Technical Implementation
Led full-stack development on a React + TypeScript + Electron stack, with LLM-assisted programming.
Translator Screenshot 1
Translator Screenshot 2
Translator Screenshot 3
Translator Screenshot 1
Translator Screenshot 2
按右键打开翻译器 CUSGA 2024 最佳解谜游戏
Right Click to Open Translator Best Puzzle, CUSGA 2024
主策划 & 谜题与叙事设计
Lead Designer & Puzzle/Narrative Design
一款围绕破译虚构语言展开的横版解谜冒险游戏。玩家在探索中逐步发现词汇和语法规则,通过“学会一门语言”本身来推进解谜与叙事。
A side-scrolling puzzle adventure built around deciphering a fictional language. Players discover vocabulary and grammar rules through exploration.
语言系统 & 谜题设计
参考自然语言学、构造语言和普遍语法理论,设计服务于谜题的虚构语言体系。核心原则:每个新语法规则的引入即是关卡的核心谜题。
竞品分析 & 迭代测试
分析同类作品确定差异化方向。每个谜题经过多轮playtest,根据卡关数据调整难度曲线与提示节奏。
主策划:范围管理与叙事协作
确定项目scope,在资源受限时做关键取舍。编写故事大纲,与文案协作打磨对话和环境叙事。
Language System & Puzzle Design
Designed a fictional language system drawing from natural linguistics and constructed languages. Core principle: the introduction of each new grammar rule IS the core puzzle.
Competitor Analysis & Iterative Testing
Analyzed comparable titles. Each puzzle went through multiple rounds of playtesting, adjusting difficulty curves based on player stuck-point data.
Lead Designer: Scope Management
Defined project scope and made critical trade-offs under constraints. Authored the story outline and collaborated on environmental storytelling.
About Me
计算机本科,游戏设计硕士(Northeastern University, Game Science & Design ’26),曾在 UC Berkeley 交换期间学习人工智能与认知科学。
设计上,我关注两个维度:一是基于玩家体验的任务与叙事设计——怎么通过节奏、情境和信息控制让玩家产生特定的情感反应;二是用数学建模和数据分析驱动的系统设计。CS背景让我能快速搭建原型并独立迭代。
我有LLM内容生成管线的实际项目经验,关注的核心问题是:当AI的行为不完全可控时,系统层面该怎么设计,才能让“不确定性”变成乐趣而不是风险。我的长期兴趣在于游戏作为一种体验媒介如何塑造人的认知与文化——以及新技术正在怎样改变这个过程。
BS in Computer Science, MS in Game Science & Design (Northeastern University ’26). Studied AI and cognitive science during an exchange at UC Berkeley.
My design work spans two dimensions: experience-driven quest and narrative design—shaping emotional responses through pacing; and analytically-driven systems design. My CS background lets me rapidly prototype and iterate independently.
I have hands-on project experience with LLM content generation pipelines. The core question I care about: when AI behavior is not fully controllable, how should we design systems to turn “uncertainty” into delight? My long-term interest lies in how games shape human cognition and culture.
2024 — 2026
Northeastern University
MS Game Science & Design
2022
UC Berkeley
交换学习 · AI与认知科学
Exchange · AI & Cognitive Science
2019 — 2024
香港中文大学(深圳)
CUHK-Shenzhen
计算机科学与技术
B.Eng Computer Science
Contact
联系方式
Get In Touch
随时欢迎交流游戏设计与合作机会
Always open to chat about game design and opportunities
WeChat QR Code
Zzzhz0817