Case Study / Details

Video Live Streaming System — Powered by AWS S3

A Japanese EdTech Company

On-Premise Professional Services Software Outsourcing

PROJECT SNAPSHOT

Video Live Streaming System

Client A Japanese EdTech Company
Industry Education
Category Software Outsourcing · Web Development · Mobile
Location Japan
Team Size 8 Professionals
Phase Coding · Unit Testing · Design
Date January 2025

OVERVIEW

Project Overview

Sonix developed a real-time video live streaming platform for a Japanese EdTech company, enabling parents to watch their children’s extracurricular activities — such as school sports events — directly from their smartphones. The system supports up to 100 simultaneous viewers per group album with end-to-end streaming latency kept under 30 seconds, while automatically generating highlight snapshots from live footage to drive photo sales revenue.

WHY SONIX

Why the Client Chose Sonix

📡

Media Streaming Expertise

Proven track record building low-latency video pipelines using MediaSoup and cloud media services, directly matching this project’s technical demands.

🌐

Bilingual Japanese–Vietnamese Team

Sonix’s bilingual engineers bridge communication gaps seamlessly, ensuring Japanese client requirements are understood and executed without distortion.

📱

Full-Stack Mobile & Cloud Capability

From Flutter-based cross-platform mobile apps to Django backends and AWS cloud infrastructure, Sonix covers the entire stack under one roof.

CHALLENGE

The Problem to Solve

The client’s existing platform suffered from low smartphone engagement — parents had little reason to open the app between photo purchase cycles. With weekly school sports events going unrecorded in real time, a significant opportunity to increase daily active users and expand the customer base was being left on the table.

Beyond engagement, the technical bar was high: the system needed to handle concurrent streams for up to 100 viewers per group album without degradation, maintain streaming delay under 30 seconds on standard mobile networks, and automatically extract usable “nice shot” images from live video — all without manual operator intervention.

Building a reliable, low-latency live streaming feature within an existing app architecture — while keeping the user experience simple enough for non-technical parents — was the core challenge.

OUR APPROACH

How We Tackled It

1

Architecture Design for Scale & Low Latency

We began by selecting MediaSoup as the WebRTC SFU (Selective Forwarding Unit) to efficiently distribute video streams to multiple viewers without re-encoding on the server, keeping CPU overhead low and latency well under the 30-second target. The architecture was designed around group-based “album” rooms, matching the client’s existing product model.

2

Automated Snapshot Extraction Pipeline

To maximise photo sales, we built an automated pipeline using AWS MediaConvert and Lambda that processes live and recorded streams asynchronously, extracts high-quality still frames at key moments, and stores them in S3 — making them immediately available for parent browsing and purchase without any manual curation effort.

3

Cross-Platform Mobile Integration & UX

Using Flutter, the team delivered a unified streaming experience across iOS and Android. The viewer UI was designed for parents in outdoor environments — large controls, clear buffering indicators, and a built-in comment feature — while the broadcaster interface remained simple enough for volunteer coaches to operate without training.

SOLUTION

What We Built

Sonix delivered a fully integrated live and recorded video streaming module built on a Python/Django backend with a Node.js signalling server powered by MediaSoup. The system handles group-scoped streaming rooms (tied to the client’s existing “album” structure), allowing up to 100 concurrent viewers per event while maintaining real-time latency below 30 seconds across standard mobile connections in Japan.

The Flutter mobile app seamlessly integrates live streaming, on-demand replay, and a contextual comment thread, while an AWS-powered post-processing pipeline (MediaConvert + Lambda + S3) automatically generates purchasable photo snapshots from video highlights. The end result is a self-running system that increases platform stickiness and opens a new revenue channel — all without requiring any manual operations from the client’s team.

TECHNOLOGIES USED

Tech Stack

PythonDjangoNode.jsMediaSoup (WebRTC SFU)FlutterAWS MediaConvertAWS S3AWS Lambda

RESULT

Outcomes & Impact

100

Concurrent Viewers per Stream

Stable group-album streams supporting up to 100 simultaneous viewers without quality degradation.

<30s

End-to-End Streaming Latency

Real-time streaming delay consistently under 30 seconds on standard Japanese mobile networks.

Auto

Photo Snapshot Generation

Zero manual curation: the pipeline automatically extracts and publishes purchasable images from every live stream.

KEY FEATURES

Feature Breakdown

Feature Description
Live Video Streaming Real-time event broadcasting via WebRTC SFU (MediaSoup), scoped to group albums with under-30-second delay.
Recorded Video Playback On-demand replay of past events stored and served from AWS S3, available immediately after the live broadcast ends.
Automatic Snapshot Generation AWS MediaConvert and Lambda pipeline extracts “nice shot” still frames from live footage and publishes them as purchasable photos without manual intervention.
Live Comment Feature Real-time comment thread on live streams, enabling parents watching remotely to engage with each other during events.

CLIENT TESTIMONIAL

Sonix delivered a stable low-latency streaming system that significantly increased parent engagement. The automatic snapshot feature also opened a new revenue stream for our platform.

— Product Manager, Japanese EdTech Company

Sonix — Your success is our happiness.

Project Details

  • A Japanese EdTech Company
  • Education
  • On-Premise Professional Services Software Outsourcing
  • 14 January 2025
  • Python Django Flutter NodeJS MediaSoup(WebRTC) AWS
  • Japan
    • Software Outsourcing
    • Integration AWS
    • Web App Application Development
    • Live Video Streaming: Real-time event broadcasting via WebRTC SFU (MediaSoup), scoped to group albums with under-30-second delay.
    • Recorded Video Playback: On-demand replay of past events stored and served from AWS S3, available immediately after the live broadcast ends.
    • Automatic Snapshot Generation: AWS MediaConvert and Lambda pipeline extracts “nice shot” still frames from live footage and publishes them as purchasable photos without manual intervention.
    • Live Comment Feature: Real-time comment thread on live streams, enabling parents watching remotely to engage with each other during events.

Contact Us For Your Next Project