Application and AI/Data Services
Domain 7 - Application and AI/Data Services. Expect product-identification questions plus design choices about where AI and data processing run.
AI services, API-first
- CLOVA Studio: build, tune and deploy generative-AI applications on NAVER's HyperCLOVA X models.
- CLOVA OCR, CLOVA Speech, CLOVA Voice: text extraction from images, speech-to-text, text-to-speech.
- Papago / NMT: machine translation.
- AI APIs are metered per call, character or image, and you remain responsible for what data you send and log.
Data and application glue
Cloud Hadoop and Data Analytics for batch and Spark processing; Data Streaming for Kafka; Search Engine for full-text search; Object Storage as the data lake. On the application side: Cloud Functions for serverless handlers, API Gateway for routing, authentication and throttling, and SNS, Push, SMS or Outbound Mailer for notifications.
Worked example
A receipt-scanning mobile feature. Reasoning: the client uploads through API Gateway to a Cloud Function; the image lands in Object Storage rather than the database; the function calls CLOVA OCR; parsed fields go to Cloud DB for MySQL; the client polls or receives an SNS notification. Serverless fits spiky traffic, and per-call OCR cost sets the budget.
Traps
- Thinking CLOVA services are models you install and run on your own Server.
- Storing images or documents as blobs in a relational database.
- Ignoring per-call quotas and cost in a high-volume pipeline.
- Assuming Cloud Functions are long-running batch workers.
- Assuming every AI service supports every language and region equally.