Senior UX professional empowering enterprise teams to ship accessible, intuitive software.
10+ years of translating dense data and legacy workflows into clear user experiences. Specializing in research, information architecture, and prototyping for Fortune 500 and federal organizations.
User Research & Usability TestingInformation ArchitectureData Visualization & DashboardsWireframing & PrototypingAccessibility · WCAG 2.2 AA / 508Agile CollaborationEnterprise Transformation
Selected work
03 case studies
Approach
Research first. Architecture always. Ship what leadership can act on.
I spend my time where consumer designers rarely do: inside dense internal platforms, regulated data, and the org politics that surround them. My work is measured in adoption, reduced friction, and decisions leaders can finally make.
U.S.-based, Agile-native, and comfortable translating design rationale for engineers, product owners, and senior leadership alike.
01
Discover
Interviews, surveys, and usability testing to separate signal from noise.
02
Architect
Sitemaps, flows, and IA that make complex systems navigable.
03
Prototype
Wireframes to high-fidelity, accessible prototypes ready for handoff.
04
Iterate
Analytics and testing drive measurable, compounding improvement.
Rebuilding CareerCircle's employer-facing sourcing & hiring platform around signal over volume, turning a "job board" into a decision-grade workspace, with accessibility built into the architecture.
Situation
Recruiters had endless applicants but no signal, sourcing meant "stabbing in the dark."
Task
Lead UX end-to-end: reframe a job board into a decision-grade sourcing workspace.
+44.6% engagement · shipped live · Gold Stevie 2026.
Focus
IA · Dashboards · A11y
Tools
Figma · Adobe XD
Timeline
2024
Context
CareerCircle (enterprise)
My Role
UX Design Lead
I led end-to-end UX: discovery research, information architecture, wireframes, and high-fidelity design, then drove design rationale with product and business stakeholders.
The Team
UX Design Lead (me)Product OwnerAccount ManagersEngineering
Enterprise hiring managers and recruiters from five partner firms joined discovery, so the platform was designed against real sourcing workflows rather than assumptions.
Scroll within the frame to explore the full screen
Overview
A centralized workspace for sourcing skilled, underrepresented talent
CareerCircle TalentConnect gives employers direct access to CareerCircle's talent community, plus tools to search candidates, build pipelines, promote jobs, and manage hiring activity in one place. This redesign evolved it from a search destination into a data-driven command center that surfaces the right candidates and the health of every requisition.
The Core Problem
Recruiters were "stabbing in the dark"
Employers weren't short on candidates, Healthcare Retailer described "thousands of applicants per req." The problem was finding the needle in the haystack. Migrations to ATS platforms like Workday had stripped away diversity visibility, making intentional diversity hiring accidental at best.
Meanwhile, fragmented workflows, hopping between LinkedIn, internal databases, and sourcing tools, created heavy cognitive load, and resume noise drained recruiter bandwidth on unqualified applicants.
Discovery & Research
I led discovery calls alongside the Product Owner and Account Managers, interviewing recruiters and hiring managers across five enterprise partners to map the recruiter journey.
Recruiters have thousands of candidates. What they lack is confidence in who to prioritize. Design for triage, not browsing.
Theme 02
Passive is the default mode
Active searching is the fallback. All four employers valued "fresh to market" digest alerts over manual Boolean searches.
Theme 03
Early-career is a blind spot
Clinical Staffing Firm and Healthcare Retailer both struggled to source near-qualified, recently-certified talent. Surface trajectory, not just seniority.
Theme 04
Respect the time budget
Solo recruiters dominate these teams. Fewer clicks beat more features; inline signal beats drill-down; email digests beat a new inbox.
Theme 05
Design for legal reality
Make source, community, and intent explicit, never infer demographics. This is what turns accidental diversity hiring into intentional.
Research Synthesis
Legacy recruiter journey map
Mapping the emotional and functional experience of enterprise partners across the legacy manual sourcing workflow, pain points tied to the phase where they surfaced.
PositivePain point
Posting
Sourcing & Search
Evaluation
Export
Print Technology Company
High manual workload
Lack of ATS integration increases posting effort.
Clinical Staffing Firm
Incremental reach
Finds early-career clinical talent not surfacing on LinkedIn.
IT Staffing Firm
“Middleman” friction
Wants jobs to auto-post directly from Connected.
Healthcare Retailer
“Stabbing in the dark”
Lost diversity visibility after the Workday transition, massive pools, no signal.
Boutique Staffing Agency
Time scarcity
Won’t sink five hours into sourcing for a single weak lead.
Clinical Staffing Firm
Resume noise
Too much time reviewing unqualified applicants without filtering.
All partners
Tool-switching fatigue
Re-identifying the same talent across disjointed systems.
Strategic Framing
The SWOT that set the UX mandate
I translated the discovery findings into a strategic SWOT that defined where design could directly mitigate business risk and drive adoption.
StrengthsS
An incremental sourcing channel outside saturated networks like LinkedIn.
Access to niche, diverse, and early-career talent pools absent from standard ATS databases.
Key insight The value prop was validated. Users wanted the data, they hated the delivery mechanism.
WeaknessesW
High cognitive load: manual Boolean searching felt like “stabbing in the dark.”
Severe resume noise: bandwidth drained reviewing unqualified applicants.
Fragmented workflows: no ATS integration forced redundant manual entry.
OpportunitiesO
Passive sourcing automation: pivot to automated digest alerts for “fresh to market” matches.
Progressive-disclosure UI: surface decision-grade signals upfront for 10-second triage.
Legal diversity visibility: ethical filters to restore diversity hiring lost in external ATS platforms.
ThreatsT
User abandonment: time-constrained recruiters abandon high-effort, low-ROI tools.
Tool-switching tax: a disconnected “middleman” destination fails enterprise adoption.
Business risk The “job board” mental model is no longer viable for enterprise teams managing high-volume reqs.
"The product had to stop forcing users to hunt for talent, pushing high-signal, actionable recommendations directly into the recruiter's existing workflow, eliminating the manual noise."
UX Strategic Mandate
The Solution
From wireframes to a decision-grade workspace
01 · Structure
Wireframing the jobs experience
I started in lo-fi to establish the information architecture, how employers view all jobs (both job-feed and manually posted), scan performance, and act, before committing to visual design.
Jobs, lo-fi wireframe
02 · The jumping-off point
A dashboard that orients, not overwhelms
A welcome moment plus at-a-glance modules, saved searches to resume passive sourcing, a jobs-health donut, pipelines, events, and accessibility resources, give recruiters a single, low-friction launch point into search, jobs, and pipelines.
Dashboard, high-fidelity
03 · Actionable insights
Job Insights, turning JPA data into sourcing strategy
Rather than a vanity dashboard, I kept reporting minimal but practical: total postings, views, and applicants, plus leaderboards for top-performing jobs, top job locations, and top job-seeker locations, so a recruiter can instantly see which requisitions need alternative sourcing.
Jobs → Job Insights, performance & leaderboards
04 · One source of truth
Job Management, every posting in one place
Employers manage job-feed and manually posted roles together, filter by source and status, and see views and applicant counts per posting, with "# New!" signals that support the passive-sourcing mental model surfaced in research.
Jobs → Job Management, feed + manual postings
05 · 10-second triage
Signal-first candidate cards
The heart of the redesign: an upfront "% match on skills and qualifications" score, validated skills as checkmarks, and explicit work-authorization signals, so recruiters make a confident decision at a glance instead of opening every profile. Diversity and community signals stay self-identified and explicit, never inferred.
Job posting → applicants, signal-first triage
Outcomes
Live
Shipped as CareerCircle’s employer platform, actively demoed and used in real enterprise engagements.
5 firms
Discovery across five enterprise partners gave every design decision traceable rationale.
By making source, community, and intent explicit rather than inferred, CareerCircle TalentConnect restored a legal, ethical path to the diversity visibility that ATS migrations had stripped away, turning accidental diversity hiring into intentional sourcing.
Also at CareerCircle
The CareerCircle.com redesign
Alongside the employer platform, I led the redesign of CareerCircle.com, the public, job-seeker-facing website that is live today. Where CareerCircle TalentConnect serves recruiters, this work served the candidates on the other side of the marketplace.
A research-driven redesign of the core job-discovery experience and the member profile-completion workflow, plus a new admin reporting and tracking suite that gave leadership real-time, data-driven visibility.
Accessibility, built in
Rather than a third-party overlay, inclusive design was embedded into the platform architecture, meeting WCAG 2.2 AA across color contrast, click targets, and screen-reader support.
+44.6%
engagement on the core job-discovery experience (48.7% to 93.7%)
+10.3
point lift in member profile-completion conversion
67→86
site health score, after resolving 140+ usability issues
Giving any CDC author a quick, easy way to create and embed consistent, interactive data visualizations directly inside the CDC's publishing system, without waiting on a designer to build them.
Situation
Scientists and writers had no fast way to add charts, so data looked inconsistent across CDC.gov.
Task
Turn a broad, complex mandate into a simple, intuitive authoring tool inside the CMS.
Action
Architected the progressive-disclosure flow & IA; Axure wireframes from single chart to dashboard.
Result
My IA became the foundation for the shipped product, the CDC's COVE.
Focus
Research · IA · Wireframes
Tools
Axure RP
Timeline
2020–2021
Context
CDC (federal)
My Role
Design & IA Lead
I owned the user research inputs, user flows, information architecture, and mid-fidelity wireframes, leading the design through architecture before rolling off ahead of launch.
The Team
Design & IA Lead (me)Business AnalystSolutions Architect3 Developers
A team of five built the data-visualization platform originally scoped for the CDC's Web Content Management System (WCMS). It later shipped publicly as the CDC's Open Visualization Editor (COVE), and comparing the live tool to my original architecture, the core flow and structure carried through largely intact. See what it became ↗
Overview
A visualization editor built into the CDC's publishing system
The platform I designed was an internal data-visualization and dashboard builder meant to live directly inside the CDC's content management system. The goal: empower scientists and researchers to quickly generate accessible, interactive visualizations for public-health articles, drastically reducing time-to-publish during the COVID-19 pandemic. It later shipped as the CDC's Open Visualization Editor (COVE).
The Core Problem · the "screenshot" bottleneck
No fast, consistent way to add a chart to an article
Scientists, doctors, and the writers publishing on their behalf had no native way to build charts inside the CDC's CMS. They built visuals in Excel or Tableau, screenshotted them, and dropped the image into the article, a slow hand-off that made every chart depend on a designer's time.
The result was visual chaos: every chart looked different depending on who made it, so data across CDC.gov was inconsistent and off-brand. Native, code-based charts would also be more accessible than a flat image, a real bonus, but the core goal was making publishing quick, easy, and consistent.
01 · Build
Chart made in Excel or Tableau
02 · Capture
Screenshot the visualization
03 · Upload
Drop a static PNG into the CMS
04 · Inconsistent
Off-brand · slow to update · not interactive
User Research & Discovery
I started with the people who had never built a chart in their lives
I partnered with research scientists, science writers, and the CDC search team to understand their technical constraints and the software mental models they already trusted. Discovery combined one-on-one interviews, a short survey, and a competitive audit of the desktop tools they were screenshotting from.
12
stakeholder interviews across science & editorial
n=40
scientists surveyed on their charting workflow
3
legacy tools audited, Excel, Tableau, Datawrapper
20+
live articles reviewed for embedded visuals
Finding 01
Not designers, not coders
Scientists reason in datasets, not SVG or HTML. The tool had to abstract the code away while keeping real control over the data.
Finding 02
Speed is survival
During the pandemic, every screenshot round-trip between a scientist and an editor cost hours the response couldn't spare.
Finding 03
508 is a gate, not a garnish
Accessibility was a legal checkpoint before publishing. Static PNGs failed it every time and blocked release.
Finding 04
Borrow a trusted model
Nearly everyone referenced Datawrapper. Matching that mental model would cut the learning curve to near zero.
Personas
Three people, one publishing pipeline
AR
Dr. Alicia Reyes
Research Scientist
Goal
Publish accurate case-trend charts within hours of finalizing the numbers.
Frustration
"I model in R and Excel, I'm not going to learn a design tool."
Needs from the tool
Paste data, pick a chart, trust it's accessible by default.
MB
Marcus Bell
Science Writer / Editor
Goal
Drop a clean, on-brand visual into an article without leaving the CMS.
Frustration
Chasing scientists for a corrected screenshot every time data shifts.
Needs from the tool
Edit data and styling inline, then re-embed with one click.
PA
Priya Anand
Section 508 Analyst
Goal
Every published asset passes accessibility review the first time.
Frustration
Static image charts are invisible to screen readers and fail audits.
Needs from the tool
Charts that ship as real, labeled markup, accessible by construction.
User Stories & Information Architecture
Structuring the tool around a progressive workflow
I translated the research into user stories, then architected a strict progressive-disclosure flow: each stage unlocks only once the previous one is valid, so scientists are never shown configuration they can't yet use.
US-01
As a research scientist, I want to upload a CSV and get a chart without writing code, so I can publish findings the same day.
US-02
As a science writer, I want to edit a chart's data and styling inline, so I don't have to request a new screenshot.
US-03
As a 508 analyst, I want visualizations to generate as accessible markup by default, so nothing static reaches production.
US-04
As a scientist, I want to group related charts into one dashboard, so a story reads as a single narrative.
Editor sitemap · progressive disclosure
Launched from the CMS article editor → "Add a visualization"
Step 01
Design Data
Import · parse · edit table
→
Step 02 · unlocks on valid data
Choose Chart
Recommended grid · live preview
→
Step 03
Configure
Axes · legend · style · table
Save to library
Reusable, versioned visualization
→
Dashboard Builder
Drag saved charts onto a canvas
→
Insert into article
Native, 508-safe embed
Mid-Fidelity Wireframes · Interactive
A strict, progressive-disclosure workflow
Step through the flow I architected, complexity is revealed only when the user is ready for it.
Step {{ cdcActive.n }} of 03
{{ cdcActive.label }}
{{ cdcActive.desc }}
→{{ note }}
{{ cdcActive.caption }}
{{ cdcStepImg }}
Designing for failure
Error handling before the chart ever breaks
Real datasets are messy. I designed explicit states for unparseable data, flagging the offending column and cell inline with a plain-language warning, so a scientist fixes the source before a broken visualization ever reaches an article.
Axure wireframe · unparseable-data error state
Scaling the System · dashboard architecture
From a single-chart editor to a full Dashboard Builder
Once the single-chart logic solidified, I led the architectural exploration for scaling the tool into a dashboard builder, mapping multiple structural options before committing engineering to one.
01 · Iterative flow mapping
Integrate, or add a dedicated step?
I evaluated whether to fold dashboard creation into the existing 3-step flow or break it into a dedicated fourth step: Design Data → Choose → Configure → Build Dashboard. Mapping each option against real system logic surfaced the tradeoffs before a line of code was written.
Flow exploration · four-step dashboard editor with loop-back to create more visualizations
02 · Modular drag-and-drop
A canvas for assembling saved visualizations
The builder introduces a modular canvas: users drag saved bar charts, maps, and line graphs from a library into a unified, responsive dashboard layout, reusing the exact configuration logic from the single-chart editor.
Dashboard Builder · drag-and-drop modules
Build Dashboard · saved-visualization library
High-Fidelity Interactive Prototype
From wireframes to a working editor
Because I rolled off before build, I reconstructed my wireframes as a high-fidelity, interactive prototype to show the experience I had designed, my envisioned finished state, not the production tool. It's styled in the CDC's U.S. Web Design System (the framework CDC.gov runs on). Load a real sample dataset, build and configure a live chart, export or embed it, then assemble saved charts into a dashboard.
Concept realization I built to demonstrate the design, not the shipped COVE product
Fully interactive · or open full screen for a larger view
Usability Testing & Optimization
Testing turned assumptions into requirements
I ran two rounds of moderated usability sessions, first on the mid-fi wireframes, then on the interactive prototype, with scientists and science writers. Each finding became a concrete design requirement handed to engineering.
What testing revealed
What changed
Users jumped straight to "Configure" before any data was loaded, then hit dead ends.
Steps 2–3 stay locked until data parses cleanly, progressive disclosure enforced in the flow.
Terms like "series" and "axis" confused non-technical scientists.
Added inline labels and a "View settings for" scoping filter so only one group of controls shows at a time.
Error messages read like stack traces; users didn't know what to fix.
Rewrote them as plain-language warnings that highlight the exact offending column and cell.
Testers didn't discover drag-and-drop when assembling a dashboard.
Added an explicit "Add to dashboard" button beside every saved chart, alongside the drag affordance.
Collaboration · Agile Delivery
Shipping the architecture with engineering
I worked in two-week sprints alongside the Business Analyst, Solutions Architect, and developers, running design reviews with scientists and the CDC search team throughout. I delivered annotated Axure specifications with acceptance criteria so engineering could build directly against the validated flow after I rolled off.
Who I partnered with
Research scientistsScience writersCDC search teamFull-stack engineersProduct owner508 program office
2-week
sprint cadence with backlog refinement
Axure
annotated specs + acceptance criteria for handoff
Outcomes & learnings
Owned the foundation
I delivered the research inputs, user flows, information architecture, and wireframes the build team carried forward.
Shipped as COVE
The platform launched publicly after I rolled off, the live tool's core flow and structure track closely to my original architecture.
Consistent by default
Native, code-based charts gave every article the same on-brand look, and, as a bonus, made the data more accessible than a flat image.
Leading the design and information architecture for a platform that shipped near-intact is the outcome I'm proudest of here. I set the direction toward native, code-based visualizations instead of static images, so charts were quick to publish and consistent across the site, and mapped the progressive-disclosure flow that scaled from a single chart to a full dashboard. Seeing that structure carry through to the live COVE product is the clearest validation the foundation was sound.
03 · Analytics Dashboards & User Research
CINC Broker Launchpad
A real-estate CRM dashboard giving brokers and team leads a bird's-eye view of their pipeline and their agents' performance, grounded in a card sort, moderated usability testing, and behavioral analytics.
Situation
Brokers and team leads had no single view of pipeline, lead sources, and team activity.
Task
Design a broker-scale analytics dashboard and validate it with real real-estate teams.
Adopted by RE teams for daily stand-ups; testers nearly unanimous they'd use it.
Focus
Research · IA · Dashboards
Tools
Sketch · InVision
Timeline
2017
Context
Commissions Inc. (CINC)
My Role
Sr. UX Designer
I ran the card sort, designed the Broker Launchpad in InVision, wrote the test plan, and co-moderated usability studies in Los Angeles and San Diego, then turned findings into design decisions.
A small CRM product team. My product owner and I recruited and moderated together; a second UX designer took notes and moderated. I owned the research plan, the prototype, and the synthesis that fed the roadmap.
The Problem
Brokers were flying blind on their team's performance
CINC's CRM gave individual agents a personal launchpad, but real-estate brokers and team leads had no equivalent bird's-eye view. Feedback from call-center reps, Intercom chats, and Net Promoter Score surveys kept pointing to the same gap.
They needed to answer three questions at a glance every morning: where are our leads coming from, how many are we generating, and what is each agent actually doing about them?
Research Process
From a card sort to moderated testing in two cities
01
Card sort
An unmoderated card sort defined which categories brokers most wanted on the dashboard.
02
InVision prototype
I designed an interactive Broker Launchpad prototype to test the IA and interactions.
03
Moderated testing
Think-aloud sessions with everyday CRM users in Los Angeles and San Diego, May 2017.
04
Behavioral analytics
FullStory session analysis to validate real in-product behavior against what users said.
2 studies · Los Angeles + San Diego12+ everyday CRM usersIn-person interview · task analysisPre / post-test surveys
Who we tested with
82%
had 5+ years of experience in real estate
80%+
work with both buyers and sellers
Daily
CINC users, lead management, communication & landing pages
DM
Dana, Team Lead
Primary persona
Goal
Start each morning knowing where leads come from, how the team is converting, and who needs coaching.
Frustration
The agent launchpad only showed one person's numbers, so running a team meant exporting spreadsheets.
Needs from the Launchpad
A single team-wide view she can scan in the daily stand-up and drill into per agent when something looks off.
Key Findings
What testing told us to keep, fix, and cut
✓ Keep
A dashboard to organize the day
Nearly all participants said they'd use the launchpad to run their day.
"I can see my team using this dashboard daily during our morning stand-up meetings to get our day started."
→ Fix
The summary bar needs the calendar
They liked the tasks-due summary, but expected it to talk to Google Calendar and email.
"I like seeing the tasks due at the top… but it would be great if this could talk to my agents' calendar in Google."
✕ Cut
"CINC Recommendations" confused users
Most found it unnecessary. I recommended removing it from the summary bar.
"What's the purpose of this? I don't need tips from CINC to do my job. Is there a way to turn it off?"
Behavioral Analytics · FullStory
The data caught what interviews couldn't
Alongside the launchpad, I used FullStory to search, watch, and analyze real user journeys in the live CRM. It surfaced a critical gap: a feature we assumed was obvious simply wasn't being found.
10%
of users knew how to assign a lead to an agent, a feature thought to be a core benefit of the app.
I used the finding to convince the product team to A/B test design fixes for the flow.
The Designs
The Broker Launchpad I designed & tested
The high-fidelity screens participants used in testing, a Today's Summary bar, lead pipeline, acquisition sources, top cities searched, team activity, and an agent overview that drills into per-agent accountability.
Broker Launchpad · main dashboard (Overview)
Agent Accountability · per-agent drill-downActivity · calls & outreach per agentLead Sources · acquisition breakdownNew Lead Trends · month over month
To Do · team tasks & reminders due
Outcomes
Validated IA
Testing confirmed the card-sorted structure matched how brokers actually run their day.
Keep · Fix · Cut
Clear, prioritized recommendations: keep the launchpad, add calendar integration, cut CINC Recommendations.
A/B testing
FullStory evidence moved the product team from opinion to experimentation on the lead-assign flow.
By pairing a card sort and moderated usability testing with behavioral analytics, the Broker Launchpad shipped with an information architecture leaders trusted, and gave the team a repeatable, evidence-led way to decide what belonged on the dashboard and what didn't.
