Unsaid

Unsaid is a minimalist, anonymous social sharing space designed for raw self-expression and confessions. The platform operates on absolute anonymity and responsive micro-interactions:

• Public Confessions Feed: A completely untraceable chronological feed featuring anonymized textual and visual cards.
• Minimalist Post Creator: An interface where users submit confessions alongside optional media assets without accounts or IP logging.
• Interactive Reactions: Engagement loops allowing users to interact with posts using pre-defined anonymous emotions.

I built the entire web app and engineered the data-anonymization architecture. Key contributions included:

• Absolute Privacy Pipeline: Decoupled post creations from user-identifiable data, stripping IP addresses, request headers, and metadata at the api endpoint before writing records.
• Optimized Media Store: Integrated client-side WebP compression and Supabase storage to process, optimize, and host confessions images under a 1MB limit for quick page load times.
• Dark Mode Aesthetic: Crafted a high-contrast dark theme with Framer Motion transitions for smooth feed interactions and content loading.

• Preventing Identity Leakage: Traditional relational schemas risk connecting posts to accounts through trace logs. I restructured the database to store user sessions in ephemeral localStorage tokens and separated authentication states from post records entirely.
• Image Load Performance: Loading raw image uploads on infinite-scroll confession feeds caused high layout shift and network latency. I resolved this by pre-compressing images to next-gen WebP formats on the client and serving them using Next.js Image optimization and BlurHash placeholders.