Safesite's legacy tools were fragmented, making safety an afterthought rather than a workflow. In partnership with product and engineering, I led an ambitious redesign of core safety workflows and built their first design system — the Beacon Design System — from the ground up.
Beacon Design System reduced handoff time by 30% through aligned component naming and interaction documentation.
Improved critical safety workflows increasing task completion rate for safety managers in the field.
Shadowing sessions with safety managers and site supervisors on construction sites, agricultural fields and warehouses.
First design system in Safesite's history — replacing 3 disparate calendar styles and inconsistent data tables.
From the start of the project, we were clear on two primary objectives: optimize the app's functionality and usability to meet and surpass internal tooling; and to shift the conversation from safety being an afterthought to being part of everyone's daily workflow — not just safety managers.
When I joined Safesite, my first instinct was to open the product and use it. What I found was a product with good bones. Core flows and ideas were sound but lacked cohesion. Buttons behaved differently across screens, spacing felt arbitrary, and patterns that should've felt familiar kept surprising me in the wrong ways.
The product felt post-MVP with lots of features shipped fast to scale quickly. Every feature made sense on its own but didn't fit well within the larger system. But before I drew a single component, I wanted to understand the people using it.
Inconsistent UI is a design problem. But whether or not those inconsistencies were actually hurting people is a question for research.
We took the research to where the product actually lived — construction sites, agricultural fields, and warehouses. Over 15+ field interviews and shadowing sessions with safety managers and site supervisors, one thing became immediately clear: customers had almost no time.
A safety manager's day is relentless. They're moving between active zones, managing workers, responding to hazards and trying to document everything in real time. The app wasn't a primary tool they sat down with. It was something they reached for to log with the little spare time afforded to them.
"This reframe changed everything. We weren't designing for users with time to think. We were designing for someone spending only minutes in the office."
Every extra tap, every extra moment of confusion, every inconsistent pattern wasn't just friction. In a safety-critical environment, it was a liability.
I synthesized qual. field data and stakeholder interviews to move beyond personas into Behavioral Archetypes. By clustering users based on their technical proficiency and psychological triggers, I identified the distinct mental models interacting with the platform.
This mapping allowed us to design a unified interface that accommodates dense data for experts while providing frictionless entry points for less technical users, reducing cognitive load and increasing reporting accuracy.
High-frequency user, deeply invested in safety outcomes. Needs speed and surface-level access to incident reporting without digging through nested menus.
Data accuracy and risk mitigation. Needs dense information displays, audit trails, and reliable export — tolerates complexity for precision.
Compliant but resistant. Uses the tool because they must. Every extra step is a reason to skip logging. Frictionless entry is critical for this user.
Power user, often in IT or operations. Manages configurations and integrations. Needs admin controls without sacrificing daily-use simplicity for others.
I also conducted a lateral research audit of dense data platforms — from fintech dashboards to logistics management tools — identifying transferable patterns that balanced information density with clarity.
To transition Safesite from a single utility tool to a complete internal platform, I redesigned the Information Architecture to support layered workflows. By auditing the existing fragmented structure, I identified that core compliance tasks were buried under deep navigational layers.
I reorganized the hierarchy into a "Hub-and-Spoke" model, ensuring that high priority actions (like Incident Reporting and Inspections) remained top level, while administrative governance lived in secondary layers.
This new IA provided the structural foundation for the Beacon Design System, allowing the platform to scale from a few core features to an enterprise-grade suite without increasing user cognitive load.
I designed these layouts to transform raw, overwhelming data into at-a-glance insights. By using clean charts and categorized modules, I made sure stakeholders could see progress without needing a manual to interpret the screen.
These wireframes weren't one-offs — they were the DNA for a scalable design system. I built these components to be modular, ensuring that as Safesite grows, the design language remains consistent and easy for developers to implement.
I also validated the wireframes in the real world — taking low-fidelity prototypes to actual safety managers and foremen. One of my favorite features looked beautiful in Figma but failed miserably when put in front of end-users. They didn't need flashy. They needed functional.
I realized that building a design system before designing the pages was a recipe for scope creep. I shifted strategy: design the core screens first to find the heart of the product, then extract those patterns to build the Beacon Design System.
This wasn't just a UI kit — it was a scalable language designed to fix years of fragmentation. I unified the experience by replacing three disparate calendar styles and inconsistent data tables with a single, robust component library in Figma.
For handoff, I moved beyond just sharing Figma links. I provided a blueprint of the why behind the what — aligning component names with the codebase, providing clear interaction documentation, and involving developers early in the exploration phase.
I treated the engineering team as my primary internal users. I involved developers early in exploration, sharing risky trendy designs to gauge technical feasibility before a single line of code was written. The best designs are the ones that actually get built.
Sitting at a desk reading analytics told me what users weren't doing. Going to construction sites told me why. You cannot design for someone spending three minutes in the office without spending a day in their shoes first.
Building a design system before understanding the product is a recipe for scope creep. Design the core screens to find the heart of the product — then extract the patterns. The system emerges; it's not imposed.
Involving developers early — sharing the risky, trendy concepts before they became specs — saved weeks of rework. The goal isn't to protect the design; it's to build the right thing. The best handoffs start at exploration, not delivery.