Chenkang Kenny Wu
Portfolio
This project focuses on improving the interdepartmental workflow of Gridware, a software company that provides utilities with the monitoring technology necessary to ensure a safe, resilient, reliable, and efficient system.
Admin V2 diminishes the significance of Admin V1. A pressing question is the current performance of V1 and the potential opportunities it might unveil. Being entirely new to the team, I was eager to embrace the challenge in collaboration.
Admin V1 addresses less than 10% of its users' workflow and hinders users from enhancing their KPI quotas.
Time is crucial, and from an initial assessment, employees face challenges in improving KPIs due to the plethora of third-party tools needed to achieve their objectives. Over-reliance on third-party apps and websites results in users having more than ten tabs open at all times.
An AI-powered device operating system that boosted inter-departmental workflow efficiencies by 55%.
A device fault and anomaly investigation system was introduced. Not only did departments seamlessly adopt the solution, but the final system also serves as a foundational platform for the company's long-term growth. Alongside the adoption of Admin V2, I began constructing a first-generation design system using atomic design principles, ensuring Admin V2 remains a sustainable solution.
What are the specific KPIs of my users and how might we improve them?
Six departments comprise the user base.
When designing an interdepartmental system, it's essential to lay a robust foundation that guarantees alignment across the cross-functional team, ensuring everyone works towards a shared objective. Here's an overview of the documents that act as the primary pillars.
Operators believe that they can work 1.5x faster and be 2x more confident when their workflow is streamlined by AI.
I was recently onboarded to this project and am prepared to tackle the challenge of crafting an exceptional solution that meets the needs of all six departments. I conducted usability interviews with over 12 senior representatives to obtain a comprehensive understanding of the current system's utility, departmental workflows, and their experiences.
A method to enhance investigation accuracy by 55%.
Monitoring device faults and anomalies brings with it the intricacies of hardware and algorithm capabilities. I encountered a study on Explainable AI Algorithms for Vibration Data-Based Fault Detection. Employing explainable AI to demystify the black-box nature of deep neural networks might be a valuable avenue to explore. By grasping the reasons behind specific classifications, users can place greater trust in the system's decisions.
"Understanding the decision process of working with a ML model is crucial to identify cases where predictions are based on incorrectly learned input-output relationships."
Users' unease with the additional clicks and their reliance on third-party platforms consistently interrupted their workflow.
With all the data at hand, I chose to process it using Affinity Mapping, Feedback Grid, and the Prioritization Matrix. These tools enabled me to grasp the high-impact actions I could undertake to develop an MVP.
The absence of an intuitive interface design led the majority of V1's users to minimize their time spent on it.
By delving into these patterns, we highlighted significant pain points and limitations, particularly regarding scalability and efficiency within the tool. For stakeholders, enhancing workflow efficiency and addressing these issues was not merely advantageous—it was essential for the project's overarching success. This evaluation solidified our direction, underscoring the need for a more flexible and efficient solution in Admin V2.
Theme 1: Investigation
With visual aids to assist in comprehending device deployment, users can rapidly grasp the magnitude of a device anomaly.
Users from different departments require varying amounts of information about a device, which also depends on the task they are undertaking.
Users feel more at ease with AI algorithm classifications when justifications are provided throughout the entire investigation process.
Theme 2: Collaboration
Cross-functional collaboration is the most essential feature for every workflows.
Team members should be able to delegate tasks to others.
With a comprehensive notification system, users are more likely to complete an investigation quickly without oversight.
Theme 3: User Interface
Operational users spend the majority of their time using the Admin Panel. They favor a minimalistic and professional interface that prioritizes functionality.
The nature of investigations necessitates actions across multiple devices, and users want the ability to access and compare different devices with minimal effort.
Employing Atomic Principles to construct components can enhance the flexibility of future features and streamline the optimization of data tables.
How can we integrate our solutions seamlessly into intricate workflows?
Creating patterns that fit once and for all.
In our team meeting, a primary objective was to analyze and comprehend the usage patterns of the current tool among different user groups. By delving into these patterns, we highlighted significant pain points and limitations, particularly regarding scalability and efficiency within the current solution.
Cross-functional collaboration and rapid iterations.
Blending brand identity with functional minimalism.
It was essential to develop a design system that not only acted as the aesthetic anchor but also embodied the core brand values we sought to communicate.
3 major improvements in design.
View Figma File
Deployment Map
A unified method to visualize fleet integrity within the viewport. This page provides a dedicated space for users to monitor and handle alerts efficiently, utilizing layers for swift data access.
Clicking on any pole opens a preview window displaying the most relevant metadata and the pole's activity history.
A robust search and filter system enables users to navigate the system with precision.
Pole View
The device page features three distinct states: the Pole View, the Alert State, and the Incident State. These states offer a unified and organized approach to device management, guaranteeing that potential problems are detected quickly, examined in-depth, and addressed efficiently through team collaboration.
This state offers a comprehensive overview of the device's status, health metrics, historical data, and other pertinent details. Users can receive real-time updates, analyze trends, and track device performance effectively.
When the system identifies anomalies using its integrated metrics, the device page shifts to the Alert State. Users can see the metrics' rationale by delving into the job history.
Moving from alerts, the Incident State signifies an elevated level of concern. It's activated when data exceeds set thresholds or when a data scientist confirms a major issue through investigation.
After multiple iterations, the Device Data Widget was refined to aid users effortlessly in accomplishing their objectives. Users can now smoothly navigate data tabs, enter pertinent parameters, and execute tasks efficiently without resorting to external applications or labor-intensive manual procedures.
Users can choose multiple items of interest from the list, such as Pole, Alerts, and Incidents. After grouping, they can carry out actions across multiple devices and review data retrieval in a single view.
Report (non disclosed)
Once the team decides to escalate an incident, stakeholders will report it to the customer. The customer reporting process begins by offering a data checklist for stakeholders to select from. The system then auto-populates the data on a page using a single-scroll design. Stakeholders can then use the language template to craft a title, description, and add customer profiles to the distribution list.
What we achieved and what's next...
30% cut in development efforts.
With the launch of our user-centric Admin V2, we seamlessly handed-off the web-based application, resulting in a 30% reduction in development efforts. This led to an impressive 55% boost in workflow efficiency. Furthermore, the overall design saw a 30% rise in user satisfaction and a 370% decrease in clicks compared to the original solution.
What are some next steps moving forward?
The Admin V2 project was a wealth of insights. Primarily, it highlighted the importance of a user-centric approach in design and development. By attentively addressing the needs and challenges of our users, we managed to provide functionalities that met their current demands and also foresaw potential future requirements.
Furthermore, collaboration stood out as a key theme – the seamless synergy among design, development, and stakeholder feedback became the cornerstone that anchored the project and guaranteed its success.
On the cutting-edge front, there's an opportunity to delve into the incorporation of emerging technologies such as AI and machine learning to further automate and enhance the anomaly detection process. In short, while we've made significant strides, the path of innovation and improvement continues.
