Capstone · UX Research & Design · Dell Technologies · Spring 2026
Making Sense of a Hardware Service System I'd Never Seen
A one-semester capstone with Dell's ISG team. I stepped into an unfamiliar world of enterprise server hardware and field service, and through research redefined what the on-product label needed to be — from a static structural map to an in-situ service support layer — then proposed a structured, multi-layer redesign.
🚀 Presented to Dell ISG label-design leadership and a cross-department roundtable. The direction is now moving toward a planned redesign.
Context
Dell's ISG servers carry System Information Labels (SILs) — the printed diagrams on the chassis that field engineers rely on to service the hardware. The visual system behind them is 10+ years old, built for an era when the label was the primary reference. I was brought in with a deceptively simple brief: modernize the label.
I had never worked in enterprise hardware, never been on a field-service call, never set foot in a data center. With one semester and a domain whose language I didn't speak, the first job wasn't to design — it was to understand, fast.
In a service workflow I'd never seen, what does the label actually need to do?
The current label: dense, lid-mounted, and built to be read away from the machine.
Research
Building a working model of an unfamiliar domain
I had about eight weeks before synthesis. I spent them triangulating three sources: working sessions with Dell's label-design lead and an industrial designer, two rounds of competitive analysis, and semi-structured interviews with two field engineers — each paired with a live concept-validation segment, where I tested early directions on the spot (arrows vs. hands, distributed vs. centralized labels, color mapping).
Onboarding→
Stakeholder sessions→
Competitive analysis→
Engineer interviews→
Concept validation→
Synthesis
The two engineers I interviewed used the system in almost opposite ways. That contrast became the backbone of my synthesis.
Engineer A
Experienced, memory-driven
Works largely from memory and treats the label as a situational backup, not a guide. Most useful for slot numbering (DIMMs) and unfamiliar layouts like risers, which differ across machines. Wants help located right where the work happens.
Engineer B
Software-first
Starts every job in digital tools and documentation; the physical label appears late in the workflow. Finds one dense sheet overwhelming, prefers distributed point-of-use references, and wants more depth for high-complexity, first-time operations.
I also studied how others solve the same problem — and looked outside the industry, at instruction systems like IKEA and LEGO, for how high complexity can stay legible.
Lenovo
Approach
Distributed, point-of-use labels
Strength
Larger, easier to read
Trade-off
Denser footprint overall
HPE
Approach
Minimal, mirrored remove/install
Strength
Low visual noise
Trade-off
Can omit needed detail
The reframe
When the label was never the tool
Across both interviews, one conclusion held: the SIL is not the primary workflow tool. Modern service runs on software diagnostics and digital documentation; the physical label is a situational support layer engineers reach for in the moment — at the rack, hands already in the machine. Designing it as a denser, prettier "complete map" would optimize the wrong thing.
So I redefined the goal. The label shouldn't try to be the manual — it should be in-situ, distributed, and contextual: support, not reference. I structured my response across three layers.
1
System — Workflow
Fit the label into the real hybrid physical–digital workflow, instead of assuming it stands alone.
2
Interaction — Physical form
Bring guidance into the physical workspace, aligned with the components engineers actually touch.
3
Visual — Representation
Make every label legible at a glance, in tight and low-light spaces.
From structural map to service support layer.
I was asked to modernize a label. The research said the label was never the real tool — so the more useful move was to redefine what it should be. That reframe, and the three-layer structure under it, was the call I drove.
Design directions
Four moves, each tied to a research insight
From the three layers, I proposed four concrete directions — a system of moves, not a single idea.
Concept
In-situ physical guidance
The boldest expression of the reframe: a semi-transparent overlay that lays directly over the open server and aligns with the real components beneath it. Instead of interpreting a diagram and counting slots, an engineer reads guidance in place — the right DIMM, the install order, the hot- vs. cold-swap touchpoints — while still seeing the hardware through it.
Rationale
Prioritize what's touched
Focus on high-frequency, high-stakes components — DIMMs, risers, fans — not every part equally.
Rationale
Color guides attention
Carry the existing hot- / cold-swap color cues onto the overlay so the meaning is reinforced, not re-learned.
Rationale
Keep the hardware visible
Semi-transparent by design — guidance never hides the structure it's describing, so engineers keep their contextual awareness of the machine.
This is a concept, not a production spec. A film laid over live boards raises real questions — static discharge, durability, whether engineers would carry it. I'd want to prototype and validate those before going further. Naming the open questions felt more honest than pretending they're solved.
Visual craft
Carrying the system down to the label itself
The same principles, applied to the labels engineers use today. I reworked two high-frequency operations — fan removal and BBU removal — through a progression: original → step-based → integrated single-view.
The space is tiny, so the goal was never more information — it was bigger, clearer information. The integrated single-view merges what were four cramped step panels into one enlarged diagram. That move drew the strongest response from Dell's team.
Principle 01
Contrast through layering
Use tonal contrast, transparency, and line weight to separate the component from the chassis and cut visual noise.
Principle 02
Scale key actions
Combine steps where it makes sense to enlarge the visuals that matter most in limited space.
Principle 03
Guide attention
Lead with arrows, highlights, and touchpoint indicators rather than hand illustrations, which engineers found ambiguous.
Principle 04
Simplify to the action
Cut non-essential detail and prioritize the actionable element, so the operational cue reads first.
BBU removal — original → step-based → integrated single-view.
Fan removal — the same progression, applied to a second high-frequency task.
Outcome
I presented the work twice — first to the label-design leadership, then in a roundtable to a wider set of the department. The reception was positive, and the direction is feeding a redesign the team is now moving forward.
Two directions drew the strongest response: the integrated single-view ("all-in-one") approach, and the strengthened visual focus that lifts active components out of the chassis.
"I'd use it more if it were easier to get to."— Field engineer, on access friction
"The help should be where my hands already are."— Field engineer, on location
"Arrows are clearer than hands for the motion."— Field engineer, on visual cues
iSchool capstone poster session, April 2026.
Reflection
I walked into a domain I knew nothing about — enterprise hardware, field service, a data center I'd never seen — and the real work was less about drawing labels than about building a model of an unfamiliar system fast enough to make a structured call. That's the part I'm proudest of.
The label was the brief. The workflow was the problem. Seeing that, and deciding to design for it, was the work.
This work is the opening move in what would be a long redesign arc. In physical products, change cycles run six months or longer — every label revision means tooling and print costs. The realistic path forward: pitch the direction internally to build small buy-in, mock it at higher fidelity, run internal user research against key metrics (task time, error rate, satisfaction), and weigh the gains against the cost of change. I understand my role was to bring a fresh lens and structured thinking to the early stage — the further this goes, the more it becomes an engineering and business conversation.
Throughout, I used AI tools to move faster — drafting concept visuals and helping organize interview notes — so more of a short timeline went to sense-making than to production.
