Creator Tools / Video Pipeline

    Onymi

    A sticker and animation creation platform. We built a high-throughput algorithm for mass processing of user videos — up to 6,000 videos per minute.

    Onymi Creator Platform•6 months
    Onymi hero

    Project Links

    App StoreGoogle Play
    RuStoresoon
    Website

    Project Overview

    Onymi operates as a high-load content generation pipeline — from raw user video to finished stickers and animations — with auto-scaling designed for peak viral traffic scenarios.

    How the high-load pipeline works

    The core of the project is a video processing pipeline: ingestion, detection, normalization, rendering, packaging, and delivery. Each stage runs as an independent, scalable service.

    • • Auto-scaling based on queue length and content type.
    • • Batch processing and optimization for high-volume user scenarios.
    • • Real-time metrics for latency, error rates, and throughput.

    Tech Stack

    Frontend · Cross-platform

    Flutter

    A single iOS and Android app built with Flutter, featuring native animations and a custom video player.

    Backend

    Self-Hosted Supabase

    Self-hosted Supabase in Docker containers on a private VPS: Auth, Realtime subscriptions, PostgreSQL, and S3-compatible storage.

    Video Processing · Microservices

    Kubernetes + Docker

    The transcoding service is built on a microservice architecture: a Kubernetes orchestrator manages Python-based Docker workers.

    Auto-scaling

    Python Workers + HPA

    Horizontal Pod Autoscaler scales Python workers under load on private VPS infrastructure — no cloud providers required.

    Project Screens

    Onymi desktop
    Main desktop screen of the Onymi platform.
    Onymi mobile screen 1
    Mobile screen for the primary user workflow.
    Onymi mobile screen 2
    Mobile screen showing user content details.