Qiaodan Ju

I’m Qiaodan Ju

We work in the darkness to serve the light.

View Details
Further Communication

About Me

about

Hello,

我是鞠桥丹，软件研发工程师，当前聚焦机器人集成部署、HMI 产品研发以及 AI 驱动研发流程落地。

I am Qiaodan Ju, a software engineer focusing on robotics deployment, HMI product engineering, and AI-driven R&D workflows.

Name: Qiaodan Ju

Birthday: 26 Feburary, 1999

Location: Shanghai, China

Email: shimmerjordan@dingtalk.com

Download CV Preview PDF Contact me

My skills

熟悉机器人中间件集成、C++、Python、ROS、UE5 引擎定制；熟练掌握自动化搭建、XXClaw 部署与优化，深度使用 AI 工具。

Skilled in robotics middleware integration, C++, Python, ROS, and UE5 customization. Strong in automation setup, XXClaw optimization, and practical AI-tool usage.

C++

95%

Python

90%

UE5

60%

RAG/AI Agent

70%

Github statistic:

Qiaodan Ju

Education

Work Experience

2025/07 - 至今 07/2025 - Present

星猿哲科技（XYZ Robotics）- MAX/HMI 研发

XYZ Robotics - MAX/HMI Engineer

星猿哲科技（XYZ Robotics）- MAX/HMI 研发（2025/07 - 至今）

  • 主导 Qt 版 HMI 的版本迭代，搭建装卸车、拆码垛一体的用户操作平台；适配 GRPC、HTTP 等多协议机械臂作业系统，接入 CSM、MAX 等自研平台。
  • 推动新版 HMI3 前后端设计与开发，使用 Vibe coding 构建 Web HMI 前端与 Python 后端；并使用 Rust 封装 Max，逐步剥离冗余模块并优化内存高位模块。
  • 主导机器人集成部署平台 MAX 研发，嵌入虚拟环境推流渲染、日志侦错回放、虚拟示教器等工具，支撑卸码一体与矩阵作业等复杂场景。
  • 搭建 HMI 与 MAX 版本迭代 CI/CD，打通需求分析到 feature 落地的 AI 全链路接入，并建设缺陷管理、冒烟问题/微小缺陷自动提单与 AI 修复系统。
  • 深度参与多现场项目定制和问题闭环，推动宝洁自动垛型码垛、小米汽车装配流水线、德国 ZARA Home 拆垛装车一体、美国 UNIS 特殊 SKU 码垛等项目部署。

XYZ Robotics - MAX/HMI Engineering (2025/07 - Present)

  • Led Qt-based HMI iteration and delivered an integrated UI platform for loading, unloading and depalletizing operations, with multi-protocol robotic arm integration (GRPC/HTTP) and in-house platform connection (CSM/MAX).
  • Drove HMI3 frontend/backend redesign using Vibe coding (Web frontend + Python backend), while refactoring Max with Rust and optimizing memory-heavy modules.
  • Led MAX deployment-platform R&D with virtual-stream rendering, log replay for debugging, and virtual teach pendant tooling for complex matrix operation scenarios.
  • Built CI/CD for HMI and MAX releases, connected AI workflows from requirement analysis to feature delivery, and introduced cloud-based defect management plus AI-assisted ticketing/fix loops.
  • Supported multi-site deployment and troubleshooting across projects including P&G, Xiaomi, ZARA Home (Germany), and UNIS (US).
2023/01 - 2025/06 01/2023 - 06/2025

华为技术有限公司 - 研发工程师 B

Huawei Technologies Co., Ltd - R&D Engineer B

华为技术有限公司 - 研发工程师 B（2023/01 – 2025/06）

  • 终端云（2024/04-2025/06）：负责多款 DAU>500w 头部手游（QQ 飞车、魂斗罗、原神等）的引擎/客户端鸿蒙化，首发完整度 95%。
  • 无线（2023/01-2024/04）：主导自动测试框架故障模式开发与 AI 驱动集成，承担可视化门户前后端与公共 API 开发。

Huawei Technologies - R&D Engineer B (2023/01 - 2025/06)

  • Terminal Cloud (2024/04-2025/06): delivered HarmonyOS adaptation for multiple top mobile games (QQ Speed, Contra, Genshin Impact, etc.), with 95% launch completeness.
  • Wireless (2023/01-2024/04): led fault-mode development and AI integration in automation testing, while building visualization portal and shared APIs.
终端云

负责多款DAU>500w头部手游（包括QQ飞车、魂斗罗、原神等）的引擎/客户端鸿蒙化，首发完整度95%

  • 完成vulkan、vsync、折叠屏开合等渲染模块适配；
  • 主导SDK（wwise、glcoud、webcamera等）及出包CICD的鸿蒙适配、重编译与集成;
  • 解决卡帧发热关键问题，设计并接入CPU负载均衡策略，达成高压测试稳定60帧，峰值温度下降7℃

    • 深度参与UTU重点模块：动画、脚本、视口渲染模块的迁移，完成AngryBot所有场景的95%资产及特效迁移

  • 设计并落地高并发导入导出框架，下沉高可拓展Unity实例序列化插件，支持十万级资产数量30s导出；
  • 负责动画状态机，解决在U3状态机的导出以及UE侧嵌套状态机、分层状态机等兼容性问题；
  • 主导UE侧U3脚本运行环境的方案设计与实现，利用IL注入实现U3属性/类的UE侧反射，优化胶水层实现以解决性能问题，最终Mono环境运行时与xlua脚本实现方案综合性能损失不超过2%；
  • 负责UE相机系统的改造，以及 CineMachine 插件及其周边脚本的UE侧适配。实现CineMachine逻辑在UE侧的100%复用以及周边模块的90%复用。在UE重建的视口渲染模块中减少技美与客户端的85%工作量。
    • 无线

    主导自动测试框架中故障模式的开发与AI驱动集成

  • 设计异步分析与冗余剪枝架构，在此框架下落地基站配置残留、脚本劣化、证书服务异常等故障模式。
  • 主导AI接入故障模式，设计Xgboost + RAG双头分析模型。基于Celery部署优先级驱动的多进程并发推理流水线，能在12h内完成工厂3w用例的失败（峰值30%失败）分析。综合节省人力50%，防线成功率与稳定性提升60%，获研究所QCC

    • 承担防线工厂的可视化门户前后端及测试框架公共API的开发

  • 承载工厂用例执行详情、防线健康监控的回显，并以版本和迭代周期为颗粒度提供入厂退厂等功能;
  • 设计并落地节电策略，端到端部署负载均衡解决方案。工作日日均节电5k度，休息日日均9k度。
  • 负责自动测试框架中公共API开发，包括云台与机械手等设备的通信与控制接口暴露。
    • Terminal Cloud

    Led HarmonyOS adaptation of engine/client for multiple top mobile games (QQ Speed, Contra, Genshin Impact, etc.), with 95% launch completeness.

  • Adapted rendering modules including Vulkan, VSync, and foldable-screen transitions.
  • Led HarmonyOS adaptation, rebuild, and integration for SDKs (wwise, glcloud, webcamera, etc.) and packaging CICD.
  • Resolved frame-drop and thermal issues by designing and integrating CPU load-balancing strategy, achieving stable 60 FPS under stress tests and reducing peak temperature by 7°C.

    • Deeply contributed to UTU core migration (animation, scripting, viewport rendering), completing migration for 95% assets and effects in AngryBot scenarios.

  • Designed and delivered a high-concurrency import/export framework with extensible Unity instance serialization plugin, supporting 100k-scale assets with 30s export.
  • Owned animation state-machine compatibility across U3 export and UE nested/layered state machines.
  • Led U3 script runtime design on UE side, implemented UE-side reflection for U3 attributes/classes via IL injection, and optimized glue layer to keep overall runtime overhead below 2% versus mono + xlua schemes.
  • Owned UE camera-system refactor and CineMachine plugin/script adaptation, achieving 100% reuse of CineMachine core logic and 90% reuse of surrounding modules, reducing 85% workload for tech-art and client teams in rebuilt UE viewport rendering module.
    • Wireless

    Led failure-mode development and AI integration in automation testing framework.

  • Designed asynchronous analysis and redundancy-pruning architecture, landing failure modes such as residual base-station configuration, script degradation, and certificate-service exceptions.
  • Led AI integration with an XGBoost + RAG dual-head model; built Celery-based priority-driven multiprocess inference pipeline that analyzed failures in 30k factory cases (up to 30% failure peak) within 12 hours, saving 50% manpower and improving success rate/stability by 60% (awarded institute QCC).

    • Developed full-stack visualization portal and shared APIs for the factory defense line.

  • Delivered execution-detail and health-monitoring visualization with version/iteration-based plant in/out capabilities.
  • Designed and deployed power-saving and end-to-end load-balancing strategies, saving ~5k kWh on workdays and ~9k kWh on rest days.
  • Developed shared APIs for device communication/control including PTZ units and robotic manipulators.

    • Education

    2021/08 - 2023/01 09/2021 - 01/2023

    新加坡国立大学 - 大数据与机器学习硕士

    Master of Science in Data Science and Machine Learning

    GPA: 4.28 / 5.0 (92.8 / 100)，TOP 7%，MERIT。

    GPA: 4.33 / 5.00 (93.3 / 100.0)

    2017/09 - 2021/06 09/2017 - 06/2021

    东北大学 - 软件工程学士

    Bachelor of Engineering in Software Engineering, Southeastern University, China

    GPA: 3.91 / 5.0 (89.1 / 100)；IELTS: 7.0；GRE: 331 + 3.5。

    GPA: 3.9199 / 5.0000 (89.199 / 100.000)

    CET-6: 610

    IELTS: 7.0

    GRE: 331 + 3.5

    Internship Experience

    05/2022 - 11/2022

    Transformer-Bumblebee 算法实习生

    Transformer-Bumblebee Algorithm Intern

    Megvii-inc, Beijing, China

  • 参与 3D 车道线检测模型研发，构建低部署压力且端到端高效的网络结构。
  • 改进预测头与后处理，设计 key-points 聚类与分段拼接策略，增强分叉车道场景鲁棒性并优化 RIM 方案。
  • 编写 curvelane/CULane 数据预处理与标注流水线，基于 MMCV 做分布式训练、坏例异步上传和版本化分析。
  • Worked on 3D lane detection to build an easy-to-deploy and efficient end-to-end network.
  • Improved prediction head and post-processing with key-point clustering and segmented lane-joining strategy for diverging lanes.
  • Built data preprocessing/annotation pipelines for curvelane and CULane, plus distributed training and bad-case async analysis tooling on MMCV.
    • 11/2021 - 03/2022

    兼职科研助理

    Part-time Research Assistance

    NUS Next++ Lab

  • 参与图生文任务研究，主要负责 backbone 重构优化、caption 评估算法设计以及高效优化器重设计。
  • 参考 DLA34 树聚合思想改进 Inception-V3，并将 Greedy Search 替换为 Beam Search，提升预测质量。
  • 重设计介于 Lazy Adam 与 Adam 之间的优化器并结合 AdamW；在 VizWiz 初步实验中获得 BLEU_1 50+、BLEU_n 约 30 的结果。
  • Contributed to image captioning research, including backbone refactoring, caption evaluation design, and adaptive optimizer redesign.
  • Improved Inception-V3 with DLA34-style aggregation and replaced Greedy Search with Beam Search for better predictions.
  • Redesigned an optimizer between Lazy Adam and Adam with AdamW; early VizWiz demo reached BLEU_1 above 50 and BLEU_n around 30.
    • 11/2020 – 02/2021

    全栈开发

    Full Stack Developer

    Neusoft Technology Co., China

    开源代码：https://github.com/shimmerjordan/sz-education

    Open-source reference: https://github.com/shimmerjordan/sz-education

  • 参与在线教学系统研发，覆盖在线课程、考试、成绩分析与管理、智能组卷、租户管理等模块。
  • 基于 SpringCloud 构建系统，集成 Consul、Spring Boot Admin、Zipkin、OAuth2/JWT、RabbitMQ 与 Redis，支撑多租户 SSO 和高并发。
  • 系统在本地 Ubuntu 20.04 部署并完成公网联通测试。
  • Built an online education system covering courses, examinations, score analysis/management, intelligent paper assembly, and multi-tenant features.
  • Implemented SpringCloud architecture with Consul, Spring Boot Admin, Zipkin, OAuth2/JWT, RabbitMQ, and Redis for multi-tenant SSO and high concurrency.
  • Packaged and deployed on Ubuntu 20.04 with public-network access testing.
    • 05/2020 – 08/2020

    全栈开发

    Full Stack Developer

    Chinasoft International, China

    项目链接：https://github.com/shimmerjordan/Lexian-Malls

    Project reference: https://github.com/shimmerjordan/Lexian-Malls

  • 参与 Lexian-Mall 生鲜批发系统研发，覆盖客户端、小程序与 CMS 后台。
  • 负责登录认证微服务下 SSO、实名模块及订单微服务购物车模块等功能开发。
  • 基于 Uni-app/Vue/SpringBoot 实现前后端能力，结合 Redis、消息队列与 Zuul 提升性能与安全，测试环境通过 10kQPS 压测。
  • Contributed to Lexian-Mall wholesale system development across client, mini-program, and CMS backend.
  • Implemented SSO, real-name authentication, shopping-cart, and order-related modules in microservice architecture.
  • Built frontend with Uni-app/Vue and backend with SpringBoot; improved performance/security via Redis, message queue, and Zuul, with 10kQPS stress-test pass.

    • Projects

    05/2026 - 05/2026

    家庭仓储语音管家（Vibe Coding 产品）

    Home Storage Voice Assistant (Vibe Coding Product)

    shimmerjordan/home_repo_git：家用收纳管理系统

    shimmerjordan/home_repo_git: a home storage management system

    • 设计并实现端到端家庭仓储系统，基于 Docker Compose 完全本地化部署。
    • 通过多 LLM provider 抽象层 + tool calling/JSON 双模式 + token-overlap 上下文摘要，实现 token 成本可控的 AI 调用。
    • 接入飞书 WebSocket Stream Mode 规避家庭 NAS 无公网环境限制，支持语音唤醒、物品位置查询及 Three.js 3D 场景高亮。
    • 实现 Function Call 需求转换（如生病需求转药物需求），并配备 SQLite 自动迁移、CSV 向后兼容、字段级审计与环形缓冲日志。
    • Designed and delivered an end-to-end home storage system with fully local deployment using Docker Compose.
    • Implemented multi-LLM provider abstraction, tool-calling/JSON dual modes, and token-overlap context summarization for cost-controlled AI usage.
    • Integrated Feishu WebSocket Stream Mode to bypass no-public-network NAS constraints, with voice wakeup, item-location query, and Three.js 3D highlighting.
    • Added Function Call intent conversion, SQLite auto-migration, backward-compatible CSV import, field-level audit logs, and ring-buffer plus rotating logs.
    10/2021 – 12/2021

    用户负债行为预测 - Kaggle 竞赛

    User Debt Behaviour Forecasting - kaggle competition

    项目链接：https://github.com/shimmerjordan/loan-pred-customer-behavior For details, please refer to the link. https://github.com/shimmerjordan/loan-pred-customer-behavior
  • 围绕非平衡数据集，分别从监督/非监督方向进行特征工程、建模预测、降维和用户聚类，目标是建立高准确度借贷行为预测模型。
  • 采用欠采样+过采样、PCA/VAE、GB/RF/XGBoost pipeline 等策略，最终获得训练集 0.97、测试集 0.89 的 F1 表现，并结合 KMeans 完成可解释性分析。
  • This unbalanced dataset was studied via supervised and unsupervised approaches, focusing on feature engineering, prediction modeling, dimensionality reduction, and user clustering.
  • Used undersampling/oversampling, PCA/VAE, and GB/RF/XGBoost pipelines; achieved 0.97 F1 on train and 0.89 F1 on test, with K-means for interpretation.
    • 09/2019 – 12/2019

    自动码垛系统

    Automated Palletizing System

    描述：实现对箱体尺寸与高度的自动识别，并通过机械臂完成杂乱箱体的有序码放。

    Description: Realized machine’s automatic identification to assess box height and size, then arranged boxes orderly via robotic arms.

    实现机械臂运动逻辑编码，使其在接收指令后可与算法模块完成请求响应闭环，从而实现自动码垛与装载。

    Implemented robot-arm motion logic so that commands can trigger request/response collaboration with algorithm modules for automatic palletizing and loading.

    10/2018 - 11/2019

    渐进式机器学习图像识别

    Image Identification Based Upon Progressive Machine Learning

    描述：将渐进式识别用于箱包图像分类，在保证准确率前提下提升分类效率。

    Description: Applied progressive identification to bag/suitcase image classification for higher efficiency with maintained accuracy.

    Wu Jianxing, Zeng Dexin, Ju Qiaodan, Chang Zixuan and Yu Hai*, “Progressive Image Recognition Method and Its Application in Security Inspection Machines”, Current Chinese Computer Science 2021; 1(1) . https://doi.org/10.2174/2665997201999201208210819

  • 参与分类网络设计与数据集构建。
  • 通过爬虫采集初始数据，结合清洗、增强与人工 GT 标注构建训练集，并通过渐进式切片模拟输入。
  • 在 VGG16 基础上融合 BN/残差思想并改进损失函数，控制不同切分比例下误分类样本惩罚因子，提升收敛效果。
  • 最终在仅输入 50% 图像的情况下，以低于 1.5% 精度损失换取约 50% 效率提升（precision 90.35%）。
  • Participated in classification-network design and dataset construction.
  • Built initial dataset via crawler scripts plus cleaning/augmentation/manual GT annotation, then simulated progressive input by segmented images.
  • Optimized VGG16 with BN/residual ideas and improved loss design to control convergence under different segmentation ratios.
  • Achieved ~50% efficiency gain with <1.5% precision loss when using only 50% image input (precision 90.35%).
    • 10/2017 - 12/2018

    RDF 焚烧发电性能研究

    Study on Electricity Generating Performance by Burning RDF

    描述：研究 RDF 不同配比下的燃烧性能与发电效率。

    Description: Studied RDF composition ratios under optimal combustion performance.

  • 分析 RDF 发电研究现状并提出优化方向。
  • 采用控制变量法制备不同混配比例的 RDF 可燃样本。
  • 基于实验统计，从失重温度与灰分分析角度评估燃烧特性。
  • Analyzed current RDF power-generation research and improvement directions.
  • Prepared combustible RDF samples with different mixture ratios via controlled variables.
  • Evaluated combustion characteristics using weight-loss temperature and ash analysis.

    • AWARDS & HONORS

    09/2017-06/2021

    Undergraduate Level

  • Second-class Scholarship for Excellent Students, NEU, China. 06/2021
  • Successful Participant, 2020 Certificate Authority Cup InternationalMATHEMATICAL CONTEST IN MODELING. 05/2021
  • Outstanding Student Pioneer at NEU, China. 04/2021
  • Second-class Scholarship for Excellent Students, NEU, China. 09/2020
  • Outstanding Student at NEU, China. 11/2019
  • Third-class Scholarship for Excellent Students, NEU, China. 10/2019
  • Third Prize at university level, The 5th “Internet +” Entrepreneurship and Innovation Competition, China. 06/2019
  • Successful Participant, Mathematical Contest in Modeling, the US. 05/2019
  • Outstanding Minster, Academy of Science and Technology, NEU, China . 04/2019
  • First-class YU Songling Scholarship, NEU, China. 11/2018
  • Third Prize at provincial level, National Mathematical Contest in Modeling for College Students, China. 10/2018
  • Second-class Scholarship for Excellent Students, NEU, China. 09/2018
  • Second Prize, Mathematical Contest for College Students, NEU, China. 07/2018
  • Seventh place in the men's single-shake jump rope category of the 57th Tohoku University Games
  • Second Prize, “PUYI” National English Composition Contest for College Students, China. 12/2017
  • 2022 NUS Distinction Honour Graduates
  • 2023 Huawei NEO Distinction Honoree; DPC Distinction Camper; Excellent Turnaround Intern; Star of Tomorrow

    • Projects

    • All Projects
    • Software Engineering
    • Machine Learning
    • Design
    • Other

    NLP

    Art, Creative
    +
    Portfolio-title

    Online-Mall

    Creative, Design
    +
    Portfolio-title

    Work Space

    Branding
    +
    Portfolio-title

    Creative Bulb

    Creative
    +
    Portfolio-title

    API

    Art, Branding
    +
    Portfolio-title

    Minimal Art

    Design, Creative
    +
    Portfolio-title
    Load more
    • 1
    • 2

    Social Works

  • Planned for the “Festival for Popularization of Science,” took charge of the debugging work of the online broadcasting program.
  • Negotiated with parts of sponsors to plan the activities, assisted in identifying the participants.

    • Academy of Science and Technology, NEU, China

    09/2019-10/2019
  • Carried on investigation on the economic development of different places in China.
  • Investigated on the poor areas in different regions, analyzed the effectiveness and expectations of targeted poverty alleviation scheme.
  • Led the team to be awarded as the “NEU Excellent Summer Practice Team”.

    • Urban and Rural Development & Targeted Poverty Alleviation Practice Team

    07/2018-08/2018

    I enjoy working with him and learn so much. You guys make the process fun and interesting. Good luck!

    customer-name

    Ying Wang

    Association Professor, NEU, China.

    Partial posts

    Go MY BLOG SITE for more details.

    Algorithm blog-title

    What is the best solution to Add Two Numbers

    • 08 December, 2020
    • Qiaodan Ju
    Machine Learning blog-title

    A Possible Way to Remove Watermark perfectly

    • 08 Novemeber, 2020
    • Qiaodan Ju
    Tech blog-title

    Comparison between Machine Learning Algorithms

    • 06 November, 2020
    • Qiaodan Ju

    Get in touch

    Phone
    Sorry, 404
    Email address
    shimmerjordan@dingtalk.com
    Location
    Sorry, 502