Projects become scenes with evidence.

Until real work is added, these are flagship slots. Later each row becomes a case study entry with motion stills, process frames, AI explorations, experience maps, and final outcomes.

Nebula OS Nebula OS

Nebula OS is the end-to-end HMI system created for the Nebula NEXT 01, a production four-door supercar developed under the Dreame - Starry Sky Program. I led the team from zero to one across the instrument cluster and center display, guided by a clear ambition: match the market in the overall experience and lead through standout modules. The result brings vehicle performance, contextual intelligence, and brand expression into one coherent cockpit experience, defining not only how the interface looks, but how the car communicates, responds, and feels in use. Nebula OS 是面向追觅-星空计划四门超级跑车 Nebula NEXT 01 打造的完整 HMI 系统。在项目从概念走向量产的过程中,我带领团队从 0 到 1 完成仪表与中控体验设计,并以“整体体验跑平,亮点模块领先”为目标,将车辆性能、情景智能与品牌表达整合为连贯统一的座舱体验。它不只定义界面如何呈现,也定义车辆如何与人沟通、响应并建立感受。

System / 01

Opening: Understanding the World Through the Screen 开篇:透过屏幕,理解世界

We live in both a physical and a digital world. Roads, vehicles, and surroundings shape the journey, while content, services, and entertainment shape digital life. They meet on the cockpit screen. Nebula OS keeps this relationship clear across driving, riding, and parked states: the desktop sustains awareness, cards interpret context, applications connect digital life, and the cluster protects essential driving information. 我们生活在物理与数字两个世界。道路、车辆与环境构成出行体验,内容、服务与娱乐构成数字生活。进入座舱,两者在屏幕上交汇。Nebula OS 在驾驶、乘坐与驻车场景中建立清晰秩序:桌面维持感知,卡片解释情境,应用连接数字生活,仪表守住驾驶信息。

01
Desktop Sustains Awareness 桌面维持感知

3D vehicle controls, wallpapers, and SR rendering have become common desktop forms for expressing technology, immersion, and personalization in smart cockpits. We organize these forms into three cognitive spaces: Drive Space presents the vehicle and its surroundings, Aura Space supports expression and mood, and Connected Space links people with vehicles and devices—helping users understand the moment before entering a task. 3D 控车、壁纸与 SR 渲染等桌面形态,已成为智能座舱强化科技感、沉浸体验与个性化表达的常见方式。我们将这些形态组织为三种认知空间:驭境呈现车辆与环境,光境承接个性与情绪,觅境连接人与设备;用户先理解此刻,再进入对应任务。

Beyond a 3D desktop that responds to day and night and an SR interface for active driving, we introduced driving-mode visualization into the 3D vehicle-control scene. Color, lighting, and motion express the performance character of each mode, while Sport+ shifts in sync with the interior color theme to deepen the immersive supercar driving experience.除随昼夜变化的 3D 桌面和驾驶状态下的 SR 界面外,我们还在 3D 控车场景中加入驾驶模式可视化,以色彩、光效与动态氛围呈现不同模式的性能特征;Sport+ 模式还会与内饰配色同步变化,进一步强化超级跑车驾驶体验的沉浸感。

02
Cards Interpret the Moment 卡片解释情境

Interviews revealed different preferences for card density, but a shared need to organize information freely by context. We atomized card content into large and small formats with flexible combinations, allowing the desktop to shift between rich detail and simplicity. In the future, voice could generate cards on demand. 访谈显示,用户对卡片的信息密度偏好不同,但共同诉求是能够按情境自由组织信息。我们将卡片内容原子化,并提供大、小两种尺寸与多种组合,让桌面在信息丰富与简洁之间灵活切换;未来还可通过语音按需生成卡片。

Nebula OS large card configurations for media, weather, energy, and tire pressure
Large card layouts / 大尺寸卡片方案
03
Applications Open Up Digital Life 应用展开数字生活

Applications support both full-screen and split-screen layouts. Users can switch freely while working and swap the left and right panels at any time. This enables focused single-task use or parallel access to navigation, settings, and controls, while a consistent visual language and interaction logic keeps every layout familiar. 应用支持全屏与分屏两种布局,用户可在使用过程中自由切换,并随时互换左右区域的位置。既能在单一任务中保持专注,也能并行查看导航、设置与控制等信息;不同布局沿用一致的视觉语言与操作逻辑,让界面随任务变化,而不增加学习成本。

VPA Exploration / 04

VPA Carries Identity Across Devices VPA 延续跨端身份

In the VPA design, beyond Dreame’s own IP characters “Bad Boy” and “Dundun Beast,” users can create a personal avatar from a photo on mobile, customize its voice, and send it to the in-vehicle system. The same VPA identity then carries across avatar selection, listening feedback, and phone-assisted task execution, making every VPA unique to its user. 在 VPA 的设计中,除了追觅本身的 IP 产品“坏小子”和“墩墩兽”,我们还可以让用户在手机端通过个人照片生成专属形象,定制声音并发送至车机。同一 VPA 身份随后延续到形象选择、聆听反馈与调用手机完成任务等状态,千人千面。

Instrument Cluster / 03

The Cluster Safeguards Driving 仪表守住驾驶

The cluster remains focused on driving decisions. Alongside familiar navigation, ADAS, and vehicle-status views, we developed multiple digital cluster styles grounded in the brand language and the character of each model. Their visual themes shift with the selected drive mode, keeping the interface aligned with the vehicle’s state and feedback. Speed, performance, energy, and alerts are reorganized by driving priority, while distinct information hierarchies and motion cues reinforce the vehicle’s personality. 仪表始终服务于驾驶判断。除了承载导航、ADAS 与车辆状态等传统驾驶信息,我们也结合品牌语言与车型特性,设计了多种数字仪表样式。仪表的视觉主题会随驾驶模式同步切换,使界面状态与车辆反馈保持一致。速度、性能、能量与告警信息按照驾驶优先级重新组织,并通过差异化的信息层级和动态反馈强化车型性格。

Appearance / 04

Between Day and Night 昼夜之间

Day and night modes adapt contrast, brightness, and surface treatment to changing cabin light without changing the information architecture or control positions. The theme shifts naturally across environments, keeping every action familiar and legible while reducing visual strain. 日间与夜间模式会随座舱光线调整对比度、亮度与界面材质,但不改变信息架构与操作位置。主题在不同环境中自然切换,让每一次操作始终熟悉、清晰,也减少明暗变化带来的视觉负担。

Nebula OS vehicle settings in day mode

System Surfaces / 05

Beyond the Desktop 桌面之外

Beyond the desktop and applications, the cockpit experience is also shaped by notifications, quick controls, device connections, climate, ambient lighting, calls, phone projection, and charging. A consistent hierarchy, surface logic, and feedback pattern connect these frequent scenarios—keeping everyday actions lightweight while allowing deeper settings to expand only when needed. 除了桌面与应用,座舱体验还由通知、快捷控制、设备连接、空调、氛围灯、来电、手机投屏与充电等系统界面共同组成。我们用一致的信息层级、浮层逻辑与反馈方式连接这些高频场景,让日常操作保持轻量,也让复杂设置只在需要时展开。

Close-up of the squared vehicle front Close-up of the rounded vehicle front

POC 概念验证

From late 2024 to mid-2025, I led a long-cycle POC for the P01 and P11 large SUVs, validating cockpit and whole-vehicle chassis capabilities before platform development. With limited software and hardware resources, we built a complete HMI solution combining live interfaces and demos across the instrument cluster, integrated center and passenger display, and rear overhead screen. Let’s continue below. 2024 年底至 2025 年中,我带领团队完成了一项持续数月的 POC 项目,围绕 P01 与 P11 两款大型 SUV,对座舱体验与整车底盘能力进行平台化车型开发前的全量验证。在软硬件资源尚不完备的条件下,我们尽可能地构建了一套完整的 HMI 解决方案,并以真实功能页面与演示视频交错呈现,覆盖仪表、中控与副驾一体屏,以及后排吸顶屏。让我们接着往下看。

First Things First

This section presents the integrated center and front-passenger display in an ultrawide 32:10 format. As the intelligent cockpit’s interaction hub, it is also the first visual focal point users encounter when they enter the vehicle. In response to real project needs, we developed three scenario-based desktop demonstrations around chassis, cockpit, and intelligent-driving capabilities. Together, they showcase whole-vehicle chassis performance, 3D vehicle controls, and an experience that integrates maps with intelligent driving.

Effective Design

At the application layer, we adopted a card-based window model to avoid a fragmented composition when apps run simultaneously across the center and front-passenger displays. At the same time, it keeps information more focused and the experience more coherent. Li Auto’s 2026 platform solution reflects a similar direction; I had already explored this approach in the ultrawide-screen concept for Great Wall Motor’s CUX 2.0 in 2021. Effective design does not chase changing forms—it solves real problems and creates a clear, coherent experience.

Super Chassis

This is a real feature. In production vehicles on the market, these functions are typically distributed across different switches in Settings. We brought them together so clients can see the chassis capabilities more directly.

01 / 09

Playful Explorations

Beyond the HMI design, the project also included vehicle-exterior showcases and a chassis-dance capability demonstration. These pieces supported content presentation on the rear overhead display while helping internal R&D engineers with vehicle tuning and functional validation.

AIGC AIGC

As the AIGC wave enters practical work, we turn this capability—from research, creation, and design to direction, automation, and delivery—into a reusable production capability, integrate it into different stages of project development, and explore multiple directions, including generating Figma-deliverable design documents from PRDs, design exploration, intelligent people-vehicle-home experience demonstrations, VPA explorations, and more. 随着 AIGC 浪潮进入实际工作,我们将这项能力从研究、创作和设计,到导演、自动化和交付,转化为可持续复用的生产能力,并融入项目推进的不同环节,围绕多个方向展开尝试,包括基于PRD生成Figma可交付的设计文档、设计探索、人车家智能体验演示、VPA 方向探索等等。

Connected Intelligence

GWM 长城

From 2019 to November 2023, I worked as a core designer on GWM's digital transformation across the CUX 1.0, 2.0, and 3.0 release cycles. The work below presents those outcomes in reverse order, from 3.0 to 1.0. 2019 年至 2023 年 11 月,我以核心设计师身份参与长城汽车的数字化转型,完整经历 CUX 1.0、2.0 与 3.0 三个主要版本周期。以下按 3.0 至 1.0 的倒序呈现阶段成果。

CUX 3.0

CUX 2.0

CUX 1.0

Hello, and I'm glad you've made it this far. I'm Xiangrui Li. I'm 31, and this is my tenth year of work. Looking back on this decade, my understanding of design, products, and myself has continued to evolve.

I started working in 2017. As I took part in more commercial projects, I came to believe that the real problem is often not a single screen, but how a product organizes information, responds to users, and can ultimately be realized by a team.

Since joining Great Wall Motor in 2019, I have worked as a core designer within the group's digital transformation strategy. I participated in three important project stages from CUX 1.0 through 3.0 and led a small team in completing vehicle-level assignments.

After joining Dreame in 2023, I continued working in HMI design across multiple proof-of-concept and production projects, while also beginning to lead teams through delivery. To me, the focus of team management is to make goals, boundaries, and standards clear; help each person contribute from their strengths; build shared judgment; and move projects forward steadily.

In recent years, AI has further changed how I work. It helps us test ideas quickly and handle repetitive work within established design standards. But the decisions that truly matter, and responsibility for the final result, still belong to people.

This portfolio does not follow the traditional PDF format. Instead, it presents the work more like a product. I hope you can explore it as you would a product: to understand my projects, and also how I observe, judge, validate, and reflect.

Looking ahead, I hope to bring these accumulated lessons into more complex products and more authentic team collaboration, creating experiences that stand up to understanding, validation, and time.

Q&A

How are innovation directions worth pursuing selected?

I do not see innovation as simply producing more ideas, but as a process of progressive convergence: first expand the pool of needs and possibilities, then evaluate them through user value, engineering feasibility, system coherence, and validation evidence, concentrating resources on the few directions that can create real product value. This approach is influenced by our company's user co-creation and “bubble-up” screening practices. The numbers change with project scale; the principle is to let valuable ideas rise through comparison and validation.

What does it mean to pursue mastery within a field?

To me, mastery is not about applying equal effort to every detail. It starts with understanding the whole problem, identifying one or two points that truly determine the experience, and developing them deeply until the result proves their value. In HMI design, that means understanding the vehicle, context, interaction, and visual system as a whole while creating a clearly above-benchmark experience in the modules that matter most.

How do you determine whether an HMI design truly works?

I evaluate it on three levels: whether it reduces cognitive and interaction cost in real driving scenarios; whether touch, voice, physical controls, visual cues, and audio feedback remain consistent across normal, exceptional, and recovery states; and whether the visual and brand expression serves the task and can be validated through a vehicle, prototype, or POC. A good HMI does more than look “high-tech”—it helps people understand faster, stay less distracted, and trust the system.