Everlum - A platform to celebrate life beautifully

Anonymous Tribute Authorization

Designing a secure system where guests can upload stories and media without accounts required a custom authorization flow. Submissions are stored in a pending state, allowing memorial owners to approve or decline content without exposing the backend to anonymous writes.

Distributed Media Compression Pipeline

To avoid expensive third‑party APIs, Everlum uses three custom microservices deployed on Render — one each for audio, images, and video. Uploads trigger a Supabase Edge Function, which forwards media to the appropriate service in a fire‑and‑forget pattern. The service compresses the file, uploads the optimized version, deletes the original, and updates the database. This system took weeks to stabilize due to Docker deployment challenges and async orchestration.

Atomic Multi‑Table Writes

Tributes, media references, and metadata span multiple tables. All writes are atomic to prevent partial submissions. Error handling was a major challenge, especially when media uploads could fail independently of database writes.

Handling Multiple Media Uploads

Visitors often upload several images or audio files at once. Everlum uses a combination of client‑side batching, server actions, and background processing to maintain performance while ensuring each file is validated, compressed, and stored correctly.

Motion Architecture with GSAP, Lenis, and Framer Motion

Cinematic motion required blending three animation systems. GSAP handles scroll‑triggered sequences, Lenis controls the global scroll feel, and Framer Motion manages component‑level transitions.

Error State Design for Older Users

Because the audience includes older demographics, error states needed to be clear, gentle, and actionable. Upload failures, validation issues, and media processing delays are surfaced with simple language and calm UI patterns.