# 唐嘉阔（Jiakuo "Torrin" Tang）

定位：AI 产品 · 用户研究 · 数据可视化（找实习）
简介：计算机本科，人机交互硕士在读；能独立完成用户调研和产品设计，借助 AI 编程工具独立开发。
可到岗：2026.11.26 起，每周 5 天｜国内城市不限
联系：jiakuotang@icloud.com｜github.com/wannian001｜网站 tangjiakuo.top
做事原则：AI 提议，人或规则决定，全程留痕

## 教育
- 墨尔本大学（THE 世界排名第 34，澳洲第 1）｜计算机科学与技术硕士（人机交互方向），2026.07～2027.12（预计）；首学期已出分的 7 项考核全部获最高档 H1：机器学习作业 96.3、软件项目管理作业 95.9、可视化作业 84.3（班均 70.0）
- 新南威尔士大学｜计算机科学与技术硕士（人工智能方向，一年后转入墨尔本大学），2025.06～2026.07；高分课程：程序设计 86（最高档 HD）；人工智能、人机交互、数据结构与算法、计算机网络均获优秀（DN）
- 吉林财经大学｜计算机科学与技术（金融信息化方向）工学学士｜GPA 3.76/5.00，2017.09～2021.06

## 项目（与简历正式版同一份文字；代码由 Claude Code / Codex 编写，需求、规格与验收由本人负责）

### AIO 主动健康管理 App（开源）｜独立负责，0→1 设计与开发，2025.10～至今
- 项目背景：睡眠、运动、饮食、屏幕时间分散在多个 App，各按大众标准打分，用户看完仍不知道今天该改什么。
- 工作内容：对比 Oura、WHOOP、Apple 健康等产品后定义需求，借助 AI 编程工具独立开发：整合手表、手机、电脑和晨间自评 4 个来源的数据，和用户自己近 28 天比，每天早上给出至多 3 条建议，次日按数据核对，多次没做到就给更小的一步。
- 项目结果：已开源并有在线演示；前身个人数字分身已自用一年、迭代 4 版，曾因数据中断 5 天没发现而加了每日数据检查。
- GitHub：github.com/wannian001/proactive-health-aio · 演示 wannian001.github.io/proactive-health-aio

### Eat It! 大学生食物管理 App｜用户研究与网页 App 开发，2026.08～至今
- 项目背景：课程课题为减少大学生食物浪费。小组对 8 名自己做饭的学生做一周日记研究和访谈，3 人认为详细记录太繁琐。
- 工作内容：负责 2 名参与者的招募、访谈和日记跟进。针对“记录太繁琐”，独立将小组方案开发成可直接使用的网页 App：做饭后用语音说一句就扣减食材，没说数量会追问；可以标出不在家吃的日子，提前提醒吃掉临期食物。
- 项目结果：网页 App 已公开可用（拍小票、扫码识别待接入）；小组调研报告 85.33/100，获最高档 H1。
- GitHub：github.com/wannian001/eat-it · 演示 wannian001.github.io/eat-it

### 维州行人事故热点地图｜数据分析与可视化（Tableau），2026.08
- 项目背景：课程作业要求为澳洲维州交通部门做决策看板，判断应优先在哪些区域改善行人安全。
- 工作内容：整合 4 张官方表、1.7 万起事故；圈选区域可按天气、光线、路况与全州对比事故占比，注明事故多可能只因人多。
- 项目结果：看板显示 65 岁以上行人占事故 20.4%，却占死亡事故 41.8%；作业获最高档 H1。
- GitHub：github.com/wannian001/vic-pedestrian-crash-map · 演示 wannian001.github.io/vic-pedestrian-crash-map

### 训练状态看板“Ready to Train?”｜数据分析（R Shiny），2026.09
- 项目背景：课程作业，为运动爱好者和教练设计看板，展示每天的状态自评、睡眠、训练和酸痛，找出与次日状态相关的习惯。
- 工作内容：用 16 人 5 个月公开数据做看板，逐一检验 16 个习惯并校正，仅“比平时晚睡 1 小时以上”与次日状态明确相关。
- 项目结果：作业 92/100，全班 291 人中最高；状态评分和人体酸痛图后来复用进 AIO 的晨间自评。
- GitHub：github.com/wannian001/ready-to-train-dashboard · 演示 wannian001.github.io/ready-to-train-dashboard

## 其他项目（简历上没有）
- 个人数字分身（个人系统）：这是我最早搭建、一直在自己用的系统：把自己的想法、生活记录和正在做的事整理在一起，让不同的 AI 助手先了解我的背景，再帮我处理任务。AIO 是后来从这套系统中拆出的健康管理应用，不是这套系统的全部。
- AI 股票研究助手（只用模拟账户，2026.09 至今）：AI 研究并提名候选名单（本人可移出），买卖只由事先定好并冻结的规则决定，AI 无权改动；只在模拟账户执行，券商连接只放行模拟盘端口。20 多项研究都先写好判定标准再看结果；2023 年数据封存到最后做一次性检验，2024 年以后的数据仍封存；已读出的结果（包括 2026.10 读过的 2023 年数据）至今无一项证明有优势，所以没有上实盘。AI 写进日报的每个数字须在原文逐字出现，否则改发程序生成的版本。
- 黑客松活动网站（墨尔本大学软件过程与管理课程，5 人，任 Product Owner，进行中）：用 MoSCoW 给 7 个用户故事排优先级、写验收标准，开发活动管理功能。
- 项目关系：个人数字分身是我最早搭建、一直在用的项目；AIO 是从它拆出的健康应用，训练后恢复是 AIO 内的场景。Ready to Train 仍是独立课程项目，其状态评分和人体酸痛图已复用到 AIO 晨间自评；它进一步完善的方向是连续跟进。Eat It! 接入 AIO 饮食管理仍为计划。

## 工作
- 中粮可口可乐饮料（吉林）有限公司｜安全生产管理专员（EHS），2023.02～2024.04：负责承包商作业的全流程安全管理：资质与入场资料审核、风险确认、作业许可、现场监督到整改落实。为推进合规，在本厂原有流程上建立成体系的检查流程，写明每步检查项和进场、关闭标准；后来直接迁入公司电子系统。独立策划、撰写脚本、拍摄并剪辑多套安全培训视频，用于集团及吉林工厂员工培训。
- 东北证券股份有限公司｜零售业务实习生，2020.07：参与数次营销活动的创意讨论，协助整理活动材料和业务表格。

## 技能
- AI 与产品：Claude Code（累计 530 亿 tokens）、Codex（累计 359 亿 tokens）写代码，我负责需求、规格和验收
- 评测与验证：自建评测题集；和简单基线对照；多重检验校正；分清关联和因果
- 研究与技术：访谈、日记研究、田野观察、亲和图、用户画像｜Python、SQL、Tableau、Figma；SwiftUI、R Shiny（AI 编写）
- 语言与其他：雅思 6.5（听力 7.0，2024.12）｜图虫签约摄影供稿人

本简介由本人维护，内容与简历正式版一致，最后更新 2026-10-21。

---

# Jiakuo "Torrin" Tang (唐嘉阔)

Looking for: an internship in AI Product · UX Research · Data Visualisation
Summary: HCI master's student at the University of Melbourne with a CS degree and 14 months in workplace safety (EHS). I take products from user research and design to working apps with AI coding tools. Available full-time in China (any city) from 26 Nov 2026.
Contact: jiakuotang@icloud.com | github.com/wannian001 | website tangjiakuo.top
Principle: the AI proposes, a person or fixed rules decide, and every step leaves a trail

## Education
- The University of Melbourne, Jul 2026 – Dec 2027 (Expected); Master of Information Technology, Human-Computer Interaction specialisation | First semester: all 7 marked assessments so far in the top band (H1), e.g. Machine Learning assignment 96.3%, Software Processes assignment 95.9%
- UNSW Sydney, Jun 2025 – Jul 2026; Master of Information Technology, Artificial Intelligence specialisation (transferred to UniMelb after one year) | Selected results: Programming 86 (HD), Data Structures & Algorithms 81 (D), Networks 81 (D), HCI 77 (D), AI 75 (D)
- Jilin University of Finance and Economics, Sep 2017 – Jun 2021; Bachelor of Engineering in Computer Science and Technology (Financial Informationisation) | GPA 3.76/5.00

## Projects (same text as the CV; Claude Code / Codex write the code, I own requirements, specs and acceptance)

### AIO Proactive Health Manager (open-source app) | Solo, 0→1 design and build, Oct 2025 – Present
- Sleep, exercise, food and screen time sit in separate apps, each scored against population norms; after looking, people still don't know what to change today.
- Compared existing products (Oura, WHOOP, Apple Health) to define the need, then built it solo with AI coding tools: it brings together four sources (watch, phone, Mac and a morning check-in), compares with the user's own last 28 days, names at most three things to adjust each morning, checks the next day whether they happened, and offers a smaller step when one keeps failing.
- Open-sourced with a live demo; grew out of my Personal Digital Twin, used daily for a year through four versions (e.g. after my data stopped for 5 days unnoticed, added a daily check that each device's data has arrived).
- GitHub: github.com/wannian001/proactive-health-aio · demo: wannian001.github.io/proactive-health-aio

### Eat It! Student Food App | User Research to Working Web App, team of 4, Aug 2026 – Present
- Course brief: reduce food waste among university students. Our team ran a one-week diary study and interviews with 8 students who cook; 3 of them found keeping track of food too much effort.
- Recruited, interviewed and followed the diaries of 2 participants. To cut that effort, built the group's design on my own into a working web app: say by voice what you cooked and it takes the ingredients off, asking how much is left when you give no amount; mark the days you eat out and it reminds you to eat what expires first.
- The web app is publicly usable (receipt and barcode recognition not yet connected); fieldwork report 85.33/100, an H1 (top band).
- GitHub: github.com/wannian001/eat-it · demo: wannian001.github.io/eat-it

### Victorian Pedestrian Crash Hotspots | Data Analysis and Visualisation (Tableau), solo, Aug 2026
- Coursework: a decision dashboard for Transport Victoria on where to improve pedestrian safety first.
- Joined 4 official tables (17,510 pedestrian crashes) into a map filterable by year, time of day and age; select any area to compare its share of crashes under weather, light and road conditions with the whole state; the page notes that more crashes may just mean more people.
- Scored 84.3/100, an H1 (top band; class mean 70.0). The dashboard shows people 65+ are 20.4% of crashes but 41.8% of fatal ones.
- GitHub: github.com/wannian001/vic-pedestrian-crash-map · demo: wannian001.github.io/vic-pedestrian-crash-map

### “Ready to Train?” Dashboard | Data Analysis (R Shiny), solo, Sep 2026
- Coursework: a dashboard for athletes and coaches showing each day's self-rated readiness (out of 10), sleep, training and soreness, and which everyday habits go with how ready people feel the next morning.
- Analysed public data on 16 people over 5 months, testing 16 habits one by one and correcting for chance links: only going to bed 1 h+ later than usual had a clear link (next-morning readiness about 0.5 points lower).
- Scored 92/100, the top mark of 291 (class mean 71.8). Its readiness rating and soreness body map were later reused in AIO's morning check-in, read with heart rate and sleep to judge recovery after training.
- GitHub: github.com/wannian001/ready-to-train-dashboard · demo: wannian001.github.io/ready-to-train-dashboard

## Other projects (not on the CV)
- Personal Digital Twin (personal system): This is the personal system I built first and still use myself. It brings together my ideas, daily records and ongoing work, so different AI assistants can understand my context before helping with a task. AIO was later split out as a health-management app; it is one part of this system, not the whole of it.
- AI Stock-Research Assistant (simulated account only, Sep 2026 – now): the AI researches and nominates a watchlist (which I can trim); rules fixed and frozen in advance, which the AI isn't allowed to change, decide buys and sells, and run only against a paper account; the broker link only opens the paper-trading port. 20-plus studies were pre-registered; 2023 data was sealed for a one-off final check (read Oct 2026, no edge either), and data from 2024 on is still sealed; no result read so far shows an edge, so nothing has gone live. Every number the AI writes into the daily brief must appear verbatim in its source, or a program-only brief goes out instead.
- Hackathon event website (Software Processes and Management, University of Melbourne; team of 5, Product Owner, in progress): prioritised 7 user stories with MoSCoW, wrote acceptance criteria, built the event-management feature.
- Project connections: the Personal Digital Twin is the long-term focus; AIO is a health application split out from it, with post-workout recovery as one use case. Ready to Train remains independent coursework; its readiness rating and soreness body map have been reused in AIO’s morning check-in, with ongoing follow-up as a direction for further design. Bringing Eat It! into AIO food management remains planned.

## Work
- EHS Specialist (Workplace Safety), COFCO Coca-Cola Beverages (Jilin) Co., Ltd., Feb 2023 – Apr 2024: Ran contractor safety end to end, from qualification checks and work permits to on-site supervision and close-out. To strengthen compliance, built the plant's loosely defined inspection process into a systematic one with every step clear (what to check, in what order, when work may start, when a job is closed); it ran on paper first, then moved straight into the company's new digital system. Independently planned, scripted, filmed and edited safety-training video series used at the Jilin plant and across the group.
- Retail Brokerage Intern, Northeast Securities Co., Ltd., Jul 2020

## Skills
- AI & Product: Claude Code (53 billion tokens to date) and Codex (35.9 billion) write the code; I own requirements, specs, task breakdown and acceptance testing
- Evaluation: self-built eval sets, simple-baseline comparisons, multiple-testing correction, telling association from causation
- UX Research & Design: interviews, diary studies, field observation, affinity mapping, personas; Figma, Tableau, R Shiny, accessibility testing (axe, WCAG)
- Technical: Python, SQL; shipped SwiftUI and R Shiny projects with AI coding tools; Git
- Languages: Mandarin (native) | English (IELTS Academic 6.5; Dec 2024)
- Other: Contracted stock photographer on Tuchong

Maintained by me; matches the current CV. Last updated 2026-10-21.