← Selected Work
AI-Guided Environment Validation
Case Study

Designing AI-Guided Environment Validation for Remote Testing

How I redesigned a high-friction room scan into a guided AI-assisted experience - reducing greeter workload, lowering candidate failure rates, and shifting environment validation from enforcement to coaching.

AI-Guided UXComputer VisionGlobalFigma
Role
Turned a high-friction four-photo room scan into a guided 360° experience
Helped move workspace validation earlier in the journey
Built rapid prototypes to test and refine the concept quickly
Type of Work
Human-AI collaboration
Rapid Prototyping (Figma)
Usability Testing
High-Level Impact
Reduced live greeter workload and helped lower check-in failure rates
Helped candidates fix environmental issues before entering the live proctoring queue
Achieved significant cost reduction while ensuring strict security compliance
Drastically Reduced
Greeter time per candidate
Globally Consistent
Workspace validation standard
Lower Failure Rate
Fewer check-in drop-offs

The friction point

AI‑guided environment validation that reduces friction before it becomes a failure.

Business context: Environment validation scans are a private, high‑stress step in remote proctoring. When guidance is unclear, candidates make preventable mistakes-and those mistakes become support cost, delays, or exam failure.

What I led: I redesigned the environment validation experience into an AI‑guided flow that coaches candidates earlier and hands off clearer evidence to reviewers.

AI‑guided environment validation that reduces friction before it becomes a failure.

What candidates were up against

Manual review created bottlenecks, inconsistency, and unnecessary anxiety.

Human‑reviewed scans were slow and inconsistent at scale. Candidates experienced the process as enforcement rather than support, increasing anxiety during a moment where trust and compliance must coexist.

My role

Lead designer partnering with product, engineering, and policy stakeholders.

  • I led the UX strategy and detailed UI for the guided scan flow.
  • I partnered with engineering to translate model outputs into understandable, actionable feedback.
  • I collaborated with compliance and operations stakeholders to ensure the experience remained policy‑accurate and globally consistent.
  • Environment validation flow screens.

Non-negotiables

Ethical AI, privacy, inclusivity, and reliable fallbacks.

  • Privacy‑sensitive environments require respectful, transparent guidance.
  • False positives and edge cases must degrade gracefully (no dead‑ends).
  • Global consistency matters-rules must be clear across regions and user contexts.

From enforcement to coaching

Shift from enforcement to prevention with coach‑like feedback.

I reframed the experience as coaching: validate earlier, explain “why,” and provide specific fixes candidates can act on. The flow is designed to reduce downstream reviewer load while improving candidate confidence.

Shift from enforcement to prevention with coach‑like feedback.

What we built

Itemized guidance + confidence handoff for trustworthy decisions.

  • Guided capture: structured the scan into clear steps with minimal ambiguity.
  • Itemized feedback: surfaced specific issues and recommended fixes, not generic errors.
  • Confidence handoff: packaged evidence so reviewers can make faster, better‑supported calls.
Itemized guidance + confidence handoff for trustworthy decisions.

What changed

Reduced operational load and improved consistency at scale.

The redesigned flow reduced time spent per candidate at the “greeter” stage and improved global consistency for workspace validation.

See the Outcomes panel for the key metrics tracked for this initiative.

What I’d do differently

AI works best as a coach-not an enforcer.

The senior design lesson was balancing accuracy with humanity: making AI feedback understandable, actionable, and fair-especially when the user is stressed.

Outcomes
Drastically Reduced
Greeter Time per Candidate
Globally Consistent
Workspace Validation