--- title: Application goals description: Define the requirements for an AI-powered event feedback backend. --- [Designing first architecture on AWS: CryptX 2.0 Hackathon Workshop](https://www.youtube.com/watch?v=OVOONoYWAjQ) The workshop starts with an event feedback system. Participants submit PDFs, audio, or video; AI services turn those files into insights that event organizers can review. The exercise focuses on backend services rather than the frontend implementation. ## What the system needs to support - **Authentication:** users can sign up and sign in securely. - **Authorization:** users and organizers can access only the operations and feedback they are allowed to see. - **Uploads:** the backend accepts several file formats without making API requests carry large files. - **Analysis:** documents and recordings are converted into content an AI model can analyze. - **Persistence:** structured results and processing metadata are stored for later queries. - **Notifications:** users or administrators learn when processing finishes. ## Separate the responsibilities Authentication belongs to Cognito, files to S3, processing to Lambda, and structured metadata to DynamoDB. API Gateway exposes the application endpoints. SQS buffers processing jobs, SNS distributes completion notifications, and Bedrock analyzes the extracted feedback. This separation lets an upload complete before the analysis finishes. It also gives each part of the backend a clear job and makes it possible to change the processing pipeline without rewriting the login or upload flow. ## Start with requirements Before choosing services, identify the expected file sizes, upload bursts, processing time, access rules, and queries organizers need. Those requirements determine whether a service is a good fit; the workshop's choices are a starting point for this particular application. [Continue to authentication and APIs](/docs/ta/workshops/aws-architecture/authentication-and-apis)