Case Studies
A selection of mock UX writing, data visualization, localization, and design projects. Each case follows a simple structure: Problem → Approach → Result.
5‑step onboarding with jargon led to a 54% completion rate and mistrust around data security.
“Complete Registration” CTA + no helper text → high drop‑off and hesitation about credit cards.
Generic error messages during checkout caused anxiety and support tickets (28% of all tickets).
The researchers developed a bite acquisition system built on a hierarchical structure with two neural networks. The first, called RetinaNet, analyzes images of plates containing different food items and places bounding boxes around each item. The second network — SPANet — uses these bounding boxes to estimate the success rates of different bite acquisition actions and to calculate the appropriate insertion angle for each item.
"We concluded that the environment surrounding the food item influences both the robot’s choice of action and its success rate. To capture this, SPANet integrates features related to environmental analysis, enabled by an environment classifier that identifies whether a food item is isolated, on the edge of a plate or another item, or stacked on top of other items.”
Bu gözlemler üzerine araştırmacılar, hiyerarşik bir yapı içerisinde iki farklı sinir ağından oluşan bir lokma alma sistemi geliştirdiler. RetinaNet isimli ilk ağ, farklı besin çeşitleri içeren dolu tabak resimlerini analiz edip her bir maddenin etrafına sınırlayıcı kutular koyuyor. İkinci ağ -SPANet- bu sınırlayıcı kutuları, farklı lokma alma eylemlerinin başarı oranını ve her madde için kullanılacak batırma açısını hesaplamak için kullanıyor.
“Besin maddesinin etrafındaki çevrenin robotun davranış seçimlerini de, başarı oranını da etkilediğine kanaat getirdik. Bu nedenle başarı oranının bulunabilmesi için SPANet'de besin maddesinin çevresi analizi ile ilgili özellikler de bulunuyor.” diye açıkladı Bhattacharjee. “Çevre özelliklerini enkode edebilmek için maddeleri üç farklı çevreden (yalnız, tabağın veya başka bir besin maddesinin kenarında, ya da başka besin maddelerinin üzerinde) birinde bulunuyor halde tanımlayabilen bir çevre sınıflandırıcısı geliştirdik.”
Field: Robotics & AI (technical subject matter)
Challenge: Translating complex terms like bounding box, insertion angle, environment classifier into precise, natural Turkish.
Strength: Balances readability with accuracy — faithful to original while sounding natural in Turkish.
Aurora Coffee Co., a boutique roaster, had a fragmented brand presence. Their logo varied across channels, colors weren’t standardized, and packaging looked disconnected from digital touchpoints. This inconsistency weakened recognition and trust, especially in competitive retail environments.
Logo: Crafted a minimal, modern logotype paired with a rising sun icon to symbolize renewal and energy.
Color Palette: Developed a warm, earthy palette — muted greens, soft beige, and deep espresso brown — to reflect natural origins and coffee culture.
Typography: Introduced a clean sans-serif primary typeface for headlines and a humanist secondary typeface for warmth in body copy.
Brand Kit: Compiled scalable guidelines covering logo usage, spacing, color codes, typography rules, and iconography.
A unified identity that seamlessly carried across packaging, website, and social media feeds.
Packaging prototypes showed stronger shelf appeal, with a consistent look that reinforced the artisanal yet modern vibe.
Social mockups demonstrated improved visual cohesion, giving the brand a professional, trustworthy presence.
Increased brand recall and recognizability in customer surveys (hypothetical for mockup purposes).
The existing website relied on outdated templates, cluttered layouts, and generic stock visuals. This caused low engagement and high bounce rates, as users struggled to quickly understand the product’s value.
Redesigned the core web visuals with a clean, modern aesthetic.
Built modular hero sections with bold typography and concise messaging.
Designed custom iconography and UI elements to match LumenTech’s brand voice.
Prioritized readability and responsive scaling, ensuring accessibility across devices.
Incorporated subtle motion cues and white space to guide the user journey.
The refreshed web visuals created an intuitive, professional look that increased user trust. Product onboarding felt smoother, while the clarified visual hierarchy improved clarity and reduced drop-offs.
Flow Fitness had scattered visuals across their social platforms. Posts lacked energy and didn’t follow a unified design language, which made the brand feel inconsistent and forgettable.
Created a set of bold, minimal templates tailored for Instagram stories, reels, and carousel posts. The system included a flexible grid, strong typography, and brand-aligned color accents. Copy tone was set to be concise, motivational, and on-brand. Templates were designed for quick adaptation to new campaign content while keeping visuals consistent.
Follower interaction increased through more engaging, dynamic visuals. The unified style gave the campaign a recognizable identity, making the feed look polished and intentional.
© 2025 Asta Studio — Istanbul, TR
info.astastudio@gmail.com