Context

This project focused on designing a BMC (Baseboard Management Controller) dashboard as a B2B product for enterprise clients.

Unlike internal tools, this system needed to support:

  • diverse engineering roles
  • different diagnostic workflows
  • cross-market expectations (Taiwan & U.S.)

The goal was not only to improve usability, but to design a system that could adapt to how different engineers interpret and act on system data.

‍

Problem

Designing the dashboard required navigating a highly complex and ambiguous system, where:

  • Critical features were not clearly defined
  • Technical terminology lacked shared understanding across teams
  • Relationships between system data and user tasks were unclear
  • Existing dashboards enforced rigid workflows that did not match real-world usage

👉 The core challenge was not just usability—
it was aligning system architecture with diverse user mental models.

‍

Research & Key Insights

To understand real-world usage, we conducted interviews with server administrators and engineers across Taiwan and the U.S.

Key Insight

Engineers do not follow a single diagnostic path.
Instead, they prioritize different signals based on experience, role, and system familiarity.

- Different users interpreted system signals differently
- Feature importance depended on task context
- Raw system data did not align with user mental models

Supporting Findings

  • Some engineers prioritize alerts, while others rely on logs or metrics
  • Senior engineers use pattern recognition; junior users require structured guidance
  • Feature importance shifts depending on task context
  • Raw system data does not align with how users think about problems

👉 This revealed a critical gap:

The system structure did not reflect how users actually diagnose issues.

‍

Approach

How might we transform complex system data into flexible, user-centered workflows?

To address this complexity, I:

  • Synthesized fragmented technical knowledge into a structured information architecture
  • Translated system-level data into user-centered groupings
  • Designed wireframes that reflect real diagnostic workflows
Step 1 — System Deconstruction
  • Reverse-engineered system functions and technical components
  • Mapped relationships between machine data and user tasks
  • Identified gaps between system logic and user understanding
Step 2 — Information Architecture (IA)
  • Synthesized fragmented knowledge into a structured IA
  • Grouped system data based on user goals and diagnostic tasks
  • Established hierarchy to prioritize critical information
Step 3 — Workflow Mapping
  • Defined multiple diagnostic paths based on user types
  • Mapped how different engineers approach system issues
  • Identified key decision points in workflows
Step 4 — Iterative Validation
  • Created wireframes reflecting real-world workflows
  • Conducted validation sessions with core users
  • Refined IA and interaction patterns based on expert feedback

‍

Key Design Decisions

Based on research and validation, I designed a configurable monitoring system that:

1. Supports Multiple Diagnostic Paths

  • Enables different users to access information based on their workflow
  • Avoids forcing a single rigid navigation structure

2. Introduces Customizable Dashboard Layouts

  • Allows engineers to prioritize the information most relevant to them
  • Supports role-based and experience-based preferences

3. Integrates Fragmented System Data

  • Combines alerts, logs, and metrics into a unified view
  • Improves visibility of critical system signals

‍

Impact

  • Reduced ambiguity in system navigation
  • Improved alignment between system data and user tasks
  • Enabled more intuitive interpretation of complex system signals

‍

Reflection

This project pushed me to think beyond UI and engage with systems at a deeper level—navigating ambiguity, technical complexity, and diverse user needs. I learned how to translate complex system logic into intuitive user workflows, and design flexible interfaces that adapt to different ways of thinking.

I’m grateful to Sylvie for the opportunity to contribute to a large-scale B2B product within a fast-paced environment. I also appreciate the guidance from my mentor, Ching, who helped me apply research methodologies in practice—especially in synthesizing complex interview data into actionable insights that shaped our design decisions. This experience strengthened my ability to navigate ambiguity and translate research into system-level design decisions.