Pawka lets pet owners scan a food product's barcode and get a nutritional score out of 100, generated by AI (GPT-4o mini) and validated by veterinarians. The project stems from a field observation: 92% of buyers cannot read pet food nutritional labels. The pet store aisle creates massive cognitive overload. Existing tools do not address this segment. Pawka solves this problem in three seconds, at the exact moment of the purchase decision. My role covers the full UX/UI process: user research (12 interviews, 3 contextual observation sessions in a pet store), persona modeling, low and high-fidelity wireframe design in Figma, atomic design system build, and validation through user testing (SUS 80, 100% success rate on the scan flow). The design system ensures consistency between the proof of concept and the prototype delivered to the Full Stack developers. WCAG 2.1 AA compliance across 14 audited criteria exceeds the standard expected for a pre-launch prototype.