yijiang.luo

Yijiang Luo 罗熠桨

Cloud & full-stack software developer. 云计算与全栈软件开发。

I build apps and the cloud services behind them. Right now that's steaL, a game-deals app with its own AWS backend. 我做应用,也做它们背后的云服务。最近在做的是 steaL,一个游戏折扣应用,后端跑在 AWS 上。

  • Monash IT, graduating Dec 2026蒙纳士大学 IT,2026 年 12 月毕业
  • Melbourne, open to roles墨尔本,正在寻找工作机会
Aussie EcoGuard ingest pipeline, which I designed. Simulated run. 我设计的 Aussie EcoGuard 处理流水线(模拟运行)
UploadBrowserpresigned URL, SHA-256 dedupe预签名 URL,SHA-256 去重
S3AWSmedia storage媒体存储
EventBridgeAWSobject-created rule对象创建事件
SQSAWSqueue + dead-letter queue队列 + 死信队列
LambdaAWScalls GCP with OAuth用 OAuth 调用 GCP
Cloud RunGCPMegaDetector, SpeciesNet, CLIP
DynamoDBAWSindexed by species按物种建索引

    

I'm an IT student at Monash University (Software Development), graduating in December 2026. I build software that runs in the cloud. I've designed a multi-cloud serverless ML platform on AWS and GCP, deployed an autoscaled API on GKE, and built full-stack web apps and native iOS and Android apps. I'm looking for software, cloud and IT consulting roles.

蒙纳士大学 IT(软件开发方向)学生,2026 年 12 月毕业。热衷于构建在云端运行的软件:做过基于 AWS + GCP 的多云 Serverless 机器学习平台、GKE 上自动扩缩容的 API,以及全栈 Web 应用和原生 iOS / 安卓应用。正在寻找软件开发、云计算及 IT 咨询相关岗位。

~/work项目

steaL

Ongoing持续开发中 Solo, Aug 2026 – now独立开发,2026.08 – 至今

Finds Steam sales, compares prices across stores, and tells you whether to buy now or wait.帮你找 Steam 折扣、跨商店比价,判断现在买还是再等。

It started as a SwiftUI iOS app and is growing into a monorepo: a native Android app, a shared Kotlin Multiplatform module and a serverless backend on AWS.最初是一个 SwiftUI iOS 应用,现在正扩展成一个 monorepo:原生安卓应用、Kotlin Multiplatform 共享模块,以及部署在 AWS 上的 Serverless 后端。

steaL home screen: hot deals with discounts and prices
steaL game detail: release date, description and the best current offer
steaL AI advice sheet recommending to wait, with a score and reasons
steaL sales calendar listing current and upcoming store-wide sales
steaL wishlist with filters and price drops
steaL profile with region, currency, theme, language and notification settings

Hot deals across stores跨商店热门折扣

Prices, history and offers for each game每款游戏的价格、史低和各商店报价

AI says buy, wait or skip, and whyAI 给出买、等还是跳过,并说明理由

Sale calendar, fed by my AWS API促销日历,数据来自我的 AWS 后端

Wishlist synced with FirebaseFirebase 实时同步的愿望单

Region, currency, theme and alerts地区、货币、主题和提醒设置

steaL/JohnWes7/steaL on GitHubGitHub 上的 JohnWes7/steaL

The original app, in SwiftUI with Liquid Glass最早的版本,SwiftUI + Liquid Glass

The same app in Jetpack Compose and Material 3. The screens are still placeholders同一个应用的 Jetpack Compose + Material 3 版本,页面目前还是占位

Kotlin Multiplatform with Ktor and Kermit. Android already uses it. iOS will once the services move in, starting with IsThereAnyDeal.Kotlin Multiplatform + Ktor + Kermit。Android 已经在用,iOS 等服务移进来后接入,第一个是 IsThereAnyDeal。

The sale calendar's API, in Python on AWS促销日历的后端接口,Python 编写,部署在 AWS

api/

openapi.yaml, the OpenAPI 3.1 contract for GET /v1/sale-eventsopenapi.yaml:GET /v1/sale-events 的 OpenAPI 3.1 契约

  • src/sale_events/The refresh logic, with no cloud SDK刷新逻辑,不依赖任何云 SDK
  • functions/Two Lambdas: the weekly refresh, and GET /v1/sale-events两个 Lambda:每周刷新,以及 GET /v1/sale-events
  • infra/Terraform for API Gateway, Lambda, S3 and the schedulerTerraform 管理 API Gateway、Lambda、S3 和定时器
  • tests/pytest on saved copies of real pages. No network, no AI, no real AWSpytest 跑在存下来的真实页面上,不联网、不调 AI、不碰真实 AWS

Every Monday it collects the store-wide sales for Steam, Epic, itch.io and GOG. DeepSeek only copies the names and dates out of each page. Code adds the times, converts to UTC and makes the IDs. Each name has to appear word for word on the page, so the AI can't make up a sale. If any check fails, the batch is thrown away and the last good data stays.每周一抓取 Steam、Epic、itch.io 和 GOG 的全站促销。DeepSeek 只负责从页面里抄出活动名和日期,时刻、UTC 换算和 id 都由代码处理。活动名必须在页面原文里逐字出现,AI 没法凭空编出一场促销。任何一项校验不通过,这次结果整份作废,保留上一份数据。

GOG has no schedule to read, so an agent searches the web and opens pages itself. Every sale it reports needs a source page and a quote, and code checks the quote is really on that page.GOG 没有可读的促销日程,所以交给一个 agent 自己搜索、打开网页。它报上来的每场促销都要附来源网页和原文引文,代码会核对引文确实出现在那一页。

Three files are AWS-specific. Moving clouds means swapping those and the Terraform.只有 3 个文件依赖 AWS,换云只需要替换它们和 Terraform。

Python / OpenAPI 3.1 / Lambda / S3 / EventBridge Scheduler / Terraform / pytest + moto

Outside the repo仓库之外

IsThereAnyDeal and the Steam Store API for prices, Firebase for accounts and the wishlist, any OpenAI-compatible LLM for adviceIsThereAnyDeal 和 Steam Store API 提供价格,Firebase 负责账号和愿望单,任意 OpenAI 兼容的 LLM 给出购买建议

Hatched parts are still being built斜纹部分还在开发中Open a folder to see what's inside点开文件夹看看里面

Agencia Change Engine

Deploying部署中 Industry project at Monash, 2026 S2蒙纳士行业项目,2026 下半年

A change-management platform for an industry client. I built the hardest part: live collaborative editing.为行业客户开发的变更管理平台。我负责团队里最难的部分:实时协同编辑。

  • Real-time co-editing with Yjs and Quill, and field-level live sync between planning tools.基于 Yjs + Quill 的实时协同编辑,多个规划工具之间字段级实时同步。
  • Token-secured WebSocket relay on a Linux server, run with Node, PM2 and Caddy.Linux 服务器上带 token 校验的 WebSocket 中继(Node、PM2、Caddy)。
  • Docker and a GitLab CI pipeline; improved the 8-step onboarding flow and wrote UAT scenarios. Shown at the Monash Industry Experience Expo 2026.Docker 和 GitLab CI 流水线;改进 8 步引导流程,编写 UAT 场景。在 Monash Industry Experience Expo 2026 展示。

CakePHP 5 / Yjs / Quill / WebSocket / Node.js / PM2 / Caddy / Docker / GitLab CI

Client project, so the source and UI are private.客户项目,源码和界面不公开。

Illustration of two people editing one plan at the same time. The client's real UI isn't shown.示意图:两个人同时编辑同一份计划。不是客户的真实界面。

Aussie EcoGuard

Technical lead, team of 3, Oct – Nov 2026三人团队技术负责人,2026.10 – 11

Tags camera-trap photos and videos by species across AWS and GCP, to track pests and find lost pets.跨 AWS 和 GCP,按物种给相机陷阱拍到的照片和视频打标签,用来追踪害兽、寻找走失宠物。

  • Designed the architecture: AWS for web, API, storage and data; GCP Cloud Run for ML inference; everything in Terraform with remote state and locking.设计整体架构:AWS 负责 Web、API、存储和数据,GCP Cloud Run 负责 ML 推理;全部资源用 Terraform 管理(remote state + locking)。
  • Compared 10 detection pipelines on a hand-labelled test set, then shipped MegaDetector, SpeciesNet and CLIP, with video support.在人工标注的测试集上对比 10 条检测流水线,最终部署 MegaDetector → SpeciesNet → CLIP,支持视频。
  • Cognito on every API; AWS calls GCP with machine-to-machine OAuth checked against JWKS, so GCP never holds AWS credentials.所有 API 走 Cognito;AWS 调用 GCP 用机器间 OAuth(JWKS 校验),GCP 侧不存 AWS 凭证。

S3 / EventBridge / SQS / Lambda / DynamoDB / Cognito / Cloud Run / Terraform / Python / pytest + moto

SmartPark

Solo, 2026 S2独立完成,2026 下半年

A FastAPI + YOLO parking-detection API on GKE that scales from 1 to 8 pods with load.部署在 GKE 上的 FastAPI + YOLO 停车位检测 API,按负载在 1–8 个 Pod 之间自动扩缩容。

  • An initContainer pulls the model weights from Cloud Storage when a pod starts. Changing the model is a ConfigMap edit.Pod 启动时由 initContainer 从 Cloud Storage 拉取模型权重,换模型只需改一下 ConfigMap。
  • Two-level cache (in-pod TTL plus shared Redis). Pods serve the stale result while one of them refreshes it under a Redis lock.两级缓存(Pod 内 TTL + 共享 Redis)。刷新期间各 Pod 照常返回旧结果,由拿到 Redis 锁的那一个 Pod 重新推理。
  • YOLO runs in a thread pool, off the event loop. Load-tested with Locust at 1, 2, 4 and 8 pods.YOLO 放在线程池里跑,不占用事件循环。用 Locust 在 1、2、4、8 个 Pod 上做了压测。

Python / FastAPI / YOLO / GKE / HPA / Cloud Run / Cloud Storage / Redis / Firestore / Locust

More work更多项目

RespiraCare patient dashboard: daily vitals tracker and the next appointment Clinician view of a patient's last 7 days of heart rate, oxygen, temperature and breathing Clinician reviewing an appointment request, with confirm and decline actions Admin dashboard with contact-message and patient-registration trends RespiraCare home page: your dedicated partner in COPD management

RespiraCare

Main developer, ~60% of commits主力开发,约 60% 提交Mar – May 20262026.03 – 05

COPD patient monitoring: patients log vitals and symptoms, doctors see 7-day trends and manage appointments.COPD 患者远程监测:患者记录生命体征和症状,医生查看 7 天趋势、管理预约。

What I built我做了什么
  • Designed the database and set up the CakePHP 5 project: authentication, password hashing, Cloudflare Turnstile.设计数据库,从零搭建 CakePHP 5 项目:认证、密码哈希、Cloudflare Turnstile。
  • Role-based access across three portals, with 404s for records you don't own.三个门户的角色访问控制,越权访问返回 404。
  • Appointment booking with HTML email notifications; password reset by time-limited 6-digit code.预约流程配 HTML 邮件通知;有时效的 6 位验证码找回密码。

PHP 8.2 / CakePHP 5 / MySQL / Tailwind CSS / ApexCharts

Stacked area chart of Australia's electricity generation by energy type, 1989 to 2023 Map of renewable energy share by state in 2023, with a year slider Small multiples of electricity generation by state, 2012 to 2023 Australia's CO2 emissions by sector, 1990 to 2023

Australia Energy & CO₂

Solo独立完成Sep – Oct 20252025.09 – 10

An interactive Vega-Lite story of how Australia's energy mix and emissions changed, from cleaned and merged official data.用清洗合并后的官方数据,以 Vega-Lite 交互图表讲述澳大利亚能源结构和碳排放的变化。

What I built我做了什么
  • Renewable-share map with a year slider (2012–2023) and a parallel-coordinates view showing coal as the main driver of emissions.带年份滑块(2012–2023)的可再生能源占比地图;平行坐标图揭示煤是排放主因。
  • Generation mix since 1989, small multiples by state, one colour system, interactive legends, light and dark modes.1989 年以来的发电结构、各州小多图;统一配色、交互图例,支持深浅色。

Vega-Lite / JavaScript / pandas / Jupyter

Traveling to Australia

Solo独立完成Jul – Sep 20252025.07 – 09

A Tableau dashboard that combines tourism and weather data so travellers can compare cities month by month.合并旅游和天气数据的 Tableau 仪表板,方便按月份对比澳大利亚各城市。

What I built我做了什么
  • Monthly temperature, humidity and rainy days, plus a compass-style wind chart built with LOD expressions and SIN/COS.月度气温、湿度、雨天数;用 LOD 表达式和 SIN/COS 做罗盘式风向图。
  • Linked legends and tooltips, weather for about 49 locations, with missing data labelled.图例联动和 tooltip,约 49 个地点的天气;缺失数据明确标注。

Tableau Public / calculated fields / LOD expressions

FallEndless title screen with start, course, options and quit Day 1 in the cave: a chest and an item to pick up, with durability and energy bars at the top left An event: a shared power bank, with three choices The event's result: durability minus 9, fuel plus 30 Day 4: interacting with an object in the cave

FallEndless

Team of 2, built in 3 days两人团队,3 天完成Apr 20212021.04

A survival game for Ludum Dare 48: pick how to respond to each day's event and manage your time and stats to last as long as you can. Built in Unity before AI coding tools.Ludum Dare 48 生存游戏:每天对事件做出选择,管理时间和能力值,尽量活得更久。在没有 AI 编程工具的年代用 Unity 完成。

Unity / C# / event, item and ending systems

~/skills技能

Cloud & infrastructure云与基础设施

AWS Lambda, API Gateway, S3, DynamoDB, SQS, EventBridge, Cognito

GCP GKE, Cloud Run, Cloud Storage, Firestore

Terraform, Docker, Kubernetes, Serverless

Languages编程语言

Python, Swift, Kotlin, C#, PHP, JavaScript, HTML/CSS

Mobile移动端

SwiftUI, SwiftData, Jetpack Compose, Material 3, Kotlin Multiplatform, Ktor

Frameworks框架与库

FastAPI, CakePHP, Node.js, Unity, Tailwind CSS, pandas, pytest, Vega-Lite

Data数据库

MySQL, DynamoDB, Cloud Firestore, Redis

Tools工具与平台

Git, GitHub, GitLab CI, Gradle, Xcode, Tableau, Jupyter

Also其他

REST and OpenAPI design, LLM apps (structured extraction, tool-calling agents), data visualisation, game developmentREST / OpenAPI 设计,LLM 应用(结构化提取、调用工具的 agent),数据可视化,游戏开发

~/education教育

Monash University

Bachelor of Information Technology, Software Development信息技术学士,软件开发方向

2025 – Dec 2026, Melbourne2025 – 2026.12,墨尔本

WAM 76.9

GPA 3.3 / 4.0

Every graded unit at Distinction or above, all core IT (FIT) units.已修课程全部为 IT 专业课,且全部达到 Distinction 及以上。

Monash units and results蒙纳士课程与成绩
Unit课程Result成绩
FIT1047 Computer Systems, Networks and SecurityHD
FIT3179 Data VisualisationHD
FIT3077 Software Engineering: Architecture and DesignHD
FIT2081 Mobile Application DevelopmentD
FIT2002 IT Project ManagementD
FIT2104 Web Database InterfaceD
FIT1049 IT Professional PracticeD
FIT3047 Industry Experience Studio 1D
FIT1050 Web FundamentalsD
FIT2175 UsabilityD
FIT3184 Cloud ComputingAwaiting result等待出分
FIT3178 iOS App DevelopmentAwaiting result等待出分
FIT3048 Industry Experience Studio 2Awaiting result等待出分

Showcased at the Monash Industry Experience Expo 2026, State Library Victoria.在维多利亚州立图书馆举办的 Monash Industry Experience Expo 2026 上展示项目。

University of Sydney

Bachelor of Engineering (Honours), prior study工程学士(荣誉),此前就读

2022 – 2023

Studied programming, object-oriented programming, data structures and algorithms, and databases before transferring to Monash.转到蒙纳士之前,修过编程、面向对象、数据结构与算法、数据库等课程。

~/contact联系

Open to software, cloud and IT consulting roles in Melbourne, on-site, hybrid or remote.正在寻找墨尔本的软件开发、云计算及 IT 咨询岗位,接受现场、混合或远程办公。

1246215651@qq.com
Work rights工作权利
Will apply for a post-study work visa after graduating.毕业后将申请毕业生工作签证。
Languages语言
Mandarin (native)
English: professional working proficiency software-specialised. Fluent in docs, error messages and code reviews; still practising the small talk.
中文(母语)
英语:专业工作水平 软件特化型。文档、报错、代码评审都看得懂,闲聊还在练。
Resume简历
Download the PDF
English, 2 pages.
中文版 PDF
英文版 PDF
各 2 页。
Elsewhere其他主页
github.com/JohnWes7
LinkedIn