AI BriefingKakao Pay news
The latest AI and developer news from Kakao Pay, with short summaries.
Visit website →Pickif(kakao)26: Shifting Focus from Conversational AI to Agentic AI
if(kakao)26 highlighted the direction of transitioning from conversational AI to execution- and trust-centric agentic AI.
KakaoPay FE Harness: Model Improvements Reduce Agents from 21 to 11
Harness size was halved due to improved model performance, but verification mechanisms must be maintained.
KakaoPay Insurance Resolves Loki OOM via Alloy Duplicate Collection Removal and Label Optimization
Removing duplicate Alloy collection and optimizing labels reduced Loki CPU usage by 98% on a sandbox basis.
Querydsl Bulk Data Processing Optimization: Applying Cursor Pagination and Batch Insert/Update
This article outlines methods to maximize bulk data read and write performance using Querydsl-based cursor pagination and batch processing.
Spring Data MongoDB Guide: Relationship Design and Update Strategies
This guide introduces strategies to resolve the N+1 problem of @DBRef using $lookup and to boost bulk update performance by 24x using bulkOps in Spring Data MongoDB.
PickKakaoPay Insurance Pursues Legacy Module Removal and Spring Boot Upgrade via AOP-Based Shadow Comparison
KakaoPay Insurance safely removed legacy modules in production and laid the groundwork for a Spring Boot upgrade using an AOP-based Shadow comparison strategy.
KakaoPay: Summary of Core Concepts and Architecture Keywords for AI Agents
A KakaoPay developer summarized essential concepts for adopting AI Agents, including the difference between LLMs and Agents, the ReAct pattern, tokens, and context windows.
Right Then, Wrong Now: A Package Manager Migration Story from Yarn Berry to pnpm
As Yarn Berry's PnP and Zero-installs became a new bottleneck, the team switched to pnpm and improved memory usage and deployment efficiency.
The journey to solve response delays occurring right after deployment (feat. JVM warm-up)
The real cause of DB Connection Timeout was not a lack of connections, but the absence of JVM warm-up.
Processing 41TB and 20 Billion Logs a Day in Real Time with ClickStack - The Hogwarts Library Project
Using OpenTelemetry and ClickHouse, log ingestion latency was cut to under 20 seconds and costs were reduced by 85.6%.
Applying FSD Architecture: Answering "Where Should This Code Go?" with Clear Code Location
Reorganizing the repo structure with FSD made code location and dependencies clear.
Building an AI Platform: From GPU Adoption to Kubeflow
They transformed manual, legacy AI operations through Kubeflow and GPU optimization.
How to Test Apps Without a Device: Multiverse Shows You How
Multiverse is an in-house platform that lets you spin up virtual devices from a single MacBook to help with app testing and log checking.
Placing a Ward on the Service: Introducing the Service Status Visualization Project 'Pink Ward'
Kakaopay Securities has introduced the 'Pink Ward' project, which visualizes service status from the customer's perspective and automates the incident response process.
A 6-Step Guide to How LLMs Work Internally for Backend Developers
This explains the process by which an LLM receives a question and generates an answer, broken down into 6 steps.
Facial Recognition and Hyper-Personalized Kiosks — Isn't a Hackathon Enough?
A team of non-developers from Kakaopay used **AWS Gen AI** to build a facial recognition-based hyper-personalized kiosk, winning 3rd place at a hackathon.
"Pageny Will Come Find You" — the Financial AI Concierge, Pageny
Kakaopay has developed 'Pageny,' a financial AI concierge that uses a multi-agent architecture based on AWS Bedrock.
My Own Loan Coach: Building a Personalized AI Loan Coaching Service
Kakao Pay's CreditClan team used AWS Bedrock to develop a personalized AI loan coach service for users.
Integrating AI Agents with KakaoPay Payment Open API: Developing the MCP Agent Toolkit
KakaoPay adopted **MCP** to efficiently connect AI agents with its payment API, and developed a multi-framework **Agent Toolkit**.
Why Do We Use Keys When Encrypting
Based on their experience refactoring an encryption module, they summarized the history of encryption and the differences between symmetric and asymmetric key methods.