AWS Enterprise Clinical Trials & Vendor Management (AWS × Johnson & Johnson)
Accelerating enterprise stakeholder decision-making efficiency by 30% through LLM-assisted research and scalable design systems.
(Note: The branding of the design assets and some of its product features has been changed to protect the confidentiality.)
Project Scope: Enterprise Analytics & AI-Driven Insights
Role: Lead UX Designer
Timeline: Dec 2025 – Feb 2026 (3 Months)
Core Tools: Figma, Qiro (LLM Data Synthesis), Lovable, Jira
Key Impact: Accelerated stakeholder decision-making efficiency by 30%
Deliverables: End-to-End UX Architecture, Journey Maps, Interactive Prototypes, Design Systems (WCAG 2.1)
The Context & Challenge
This was a high-stakes, 3-month enterprise proposal project for Johnson & Johnson. Data Acquisition Experts (DAEs) were drowning in manual, fragmented data silos when trying to identify, evaluate, and assign clinical trial vendors.
The Core Bottlenecks:
Fragmented data across multi-source EHR and lab feeds causing severe decision-making delays.
Zero centralized visibility into vendor performance metrics or trial anomalies.
High operational costs and strict compliance constraints (WCAG & healthcare standards).
The Problem Architecture Diagram
AI-Powered Research & SME Discovery
Due to strict time constraints and zero direct access to end-users, we leveraged Subject Matter Experts (SMEs) alongside LLM-driven data synthesis (Qiro) to parse raw transcript logs and discovery notes.
LLM Data Extraction: Fed qualitative SME interview transcripts and legacy workflow logs into LLMs to instantly extract core structural requirements and cluster pain points.
Journey Mapping: Developed a unified DAE journey map to pinpoint exact operational drop-off points during vendor coordination.
Execution: Architecture, Systems & AI Prototyping
Design Systems: Leveraged the AWS Cloudscape Design System paired with custom tokens to ensure immediate enterprise scalability and WCAG accessibility compliance.
Generative UI & Exploration: Used generative workflows to rapidly blueprint structural variations and edge-case "unhappy paths" before high-fidelity execution.
AI-Assisted Prototyping (Lovable): Transitioned exploratory design concepts into high-fidelity, functional web applications using Lovable, shortening design-to-prototype cycles by 25% and validating live data handling prior to engineering handoff.
Results & Impact
30% Efficiency Boost: Accelerated stakeholder decision-making speed through streamlined data dashboards.
Client Validation: The prototype was perceived by Johnson & Johnson stakeholders as a fully functional product rather than a static concept.
Key Constraints & Learnings
Domain Complexity: Navigated specialized clinical trial workflows by relying on rigorous SME validation loops.
The Constraint of Remote Collaboration: Overcame the lack of direct user access by building robust simulation models via AI-assisted full-stack prototyping tools.