Kakao Hosts Third Internal Tech Seminar 'Techtalk' Covering Cloud, ML, and Elasticsearch

Kakao hosted the third event of its internal technical seminar Techtalk, sharing technical experiences on three topics: cloud, ML/DL, and Elasticsearch. This seminar was organized to facilitate knowledge exchange and refreshment for developers in a remote work environment, with some content planned for external release in the future.

Cloud Infrastructure Automation and Migration

The Krane team, responsible for Kakao's cloud IaaS service, introduced the process of transitioning to a container deployment environment using existing abstracted deployment code. The session detailed attempts to safely migrate massive infrastructure and complex domain knowledge without significant changes or failures, as well as the evolution of the CI/CD system used to validate these efforts.

Real-Service Application Cases for ML/DL Projects

A case study on the introduction of ML/DL for a news comment safety-related project was presented. It covered the journey from the prototyping stage to actual service deployment, sharing the challenges faced by server-centric organizations when adopting machine learning and the operational experience post-deployment. This is expected to serve as a practical reference for organizations aiming to scale beyond side projects to real services.

Elasticsearch Performance and Cost Optimization

An engineer responsible for advertising services presented methods for optimizing indexing and search performance in Elasticsearch. Emphasizing the significant difference between basic usage and professional utilization, the session explained core principles of indexing and search that are often overlooked, along with efficient usage techniques to resolve issues of performance degradation and cost increases.