A Content Creation Platform for Talent and Advertisers

    We were tasked with the development of TLYNT, a dual-faceted content creation platform aimed at two sets of users: content creators and advertisers. The platform is designed to facilitate a seamless flow of project-based content creation and approval. The Talent uses a mobile app to join projects, upload videos, and track project status, while Advertisers use a web app for project creation, video review, approval, and export. The entire system is built on AWS services, offering scalability, reliability, and performance.

    Challenges

    Dual User-Base: Providing an intuitive, performant experience for two types of users—Talent and Advertisers.

    Async Flow and Updates: Ensuring near real-time status updates for asynchronous video transcoding and other processes without relying on long running blocking requests.

    Media Handling: Video transcoding and optimization to cater to various formats and devices, advertisers have to review a significant amount of videos quickly and loading time has a huge impact on the user experience.

    Resource Scalability and Availability: The media processing pipeline had to be able to handle a lot of requests at once without having a slow increase beforehand.

    Service Architecture

    The TLYNT platform is architected to leverage a variety of AWS services for optimal performance and scalability:

    GraphQL and REST APIs: Deployed as ECS containers for optimal performance and scalability, they communicate with each other directly thanks to Service Connect within the space of the ECS cluster to reduce network latency.

    S3 Buckets: Used for storing both media files and the web app resources. CloudFront: CDN for web app and media distribution, enhancing load speed and performance thanks to adapted caching based on content type and path.

    Lambdas and SQS: Handle asynchronous tasks such as media transcoding and archive creation, offering event-driven scalability.

    AWS Elemental MediaConvert: Used to transcode the original file uploaded by the user to a stream optimized format.

    For simplicity, availability zones and replication have been omitted from the diagram.

    The Media Flow

    1. Talent App requests to upload a video

    2. API generates a multi-part upload link for the media S3 bucket

    3. Talent App uploads the video to S3 without the multi-part link

    4. Newly uploaded video triggers an event that is dispatched to SQS

    5. Media Processing Lambda consumes the SQS queue, create an Element MediaConvert job, and update the media processing status

    6. Elemental MediaConvert dispatches job events to SQS

    7. Media Processing Lambda consumes the SQS queue and update the media processing status

    Conclusion

    Leveraging event-based, asynchronous computing coupled with AWS Elemental MediaConvert allowed us to create a service with zero cost when there is no media to process, and that is able to scale to accommodate a sudden volume of requests, with asynchronous processing and parallelization.

    Westfield Line Pass

    Tech Holding builds a line pass mobile experience to help retailers adhere to COVID guidelines.

    Tech Holding builds a line pass mobile experience to help retailers adhere to COVID guidelines.

    By using this site, you agree to thePrivacy Policy.