God of Highschool Game Testing

Table of Contents Toggle OverviewChallenges & SolutionsChallenge 1: The “Legacy” Codebase FactorChallenge 2: Market Suitability for Modern SEA GamersChallenge 3: Extreme Sprint TimelineScope of Work & Team StructureBusiness Impact Overview Game Genre Gacha Turn-based RPG based on a famous South Korean Webtoon/Manhwa Core Technology Strategic team-building, elemental counters, and special character abilities (“Charyeok”) Platforms Published by Indonesia’s leading media group, targeting the Indonesian and wider Southeast Asian (SEA) mobile market. Challenges & Solutions Challenge 1: The “Legacy” Codebase Factor The Problem: The original game has been live globally for over a decade. Its source code contains 10 years of layered updates, making it highly likely that a newly localized version for Indonesia would inherit old technical debt, performance issues, or hidden legacy bugs. Our Solution: We conducted a deep Code & Feature Regression Sweep, prioritizing the core gameplay loop and oldest system modules to catch hidden bugs early.We ran intensive long-session stability tests to monitor memory leaks, ensuring the older architecture runs smoothly on modern mobile OS versions. Challenge 2: Market Suitability for Modern SEA Gamers The Problem: A game that succeeded globally 10 years ago might not appeal to modern Indonesian and SEA mobile players. The publisher needed real data on whether the progression speed, gacha drop rates, and difficulty curves would still resonate with today’s market. Our Solution: We formed a dedicated Playtest Focus Group consisting of experienced QA players who specialize in modern anime/manga gacha games in the SEA region.Beside standard functional testing, our team playtested competing top-tier gacha games currently active in the market. We delivered a comprehensive Competitive Benchmark Report to the publisher, providing detailed feedback on what the game does well and actionable advice on adjusting drop rates, reward systems, and difficulty curves to improve player retention Challenge 3: Extreme Sprint Timeline The Problem: The publisher required five specialized types of QA (Functional, LQA, Compatibility, Regression, and Market Playtesting) to be completely finished in under three weeks, which demanded an incredibly high level of efficiency. Our Solution: We used a Dynamic Resource Allocation strategy, splitting the QA team into concurrent sub-teams that worked in parallel.While the functional team verified system logic, the LQA and Playtest teams worked simultaneously inside optimized test environments, allowing us to deliver all five testing phases successfully before the strict 3-week deadline Scope of Work & Team Structure Scope of Work Informed Go/No-Go Decision: Provided the MNC Group with a data-driven “Market Appeal” report, allowing them to evaluate the project’s ROI before a full-scale marketing blitz. Modernized Stability: Identified and cleared legacy bugs that would have hindered the experience on modern OS versions (Android 14/iOS 17+). Balanced Economy: Verified that the game’s monetization and difficulty curves were fair and sustainable for the target audience, protecting long-term retention. Timeline Achievement: Delivered a 360-degree quality assessment within the strict 18-day window, enabling the client to hit their strategic market-timing goals. Team Structure 1 QA Leader 10 QA Engineers Business Impact All Features are tested with the release on time and high quality without the blocker issue For Market Analysis QA and Balancing QA, the team delivered results that met the client’s expectations. The Fail rate is 0.41% for Functional QA and 2.59% for Compatibility QA The total project effort delivered value exceeded expectations compared to the initial budget plan

Table of Contents Toggle OVERVIEWSCOPE OF WORKAI SolutionSOLUTIONS & STRATEGIESRESULTS AND IMPACTS – ARCHITECTUAL PATTERN OVERVIEW Our client, whose expertise is in the field of LLM, needed a system to evaluate and optimize Retrieval-Augmented Generation (RAG) pipelines for their specific data and use cases. The goal was to automate the process of testing various RAG modules and configurations to find the optimal pipeline for their unique requirements. Country: Korea Domain: Technology Services AI expertise: LLM Development process: Scrum SCOPE OF WORK RAG Pipeline Evaluation Modular RAG Components Automated Testing Cloud Integration Performance Metrics Auto-deployment to chat AI Solution Created a modular RAG pipeline architecture supporting various components Implemented comprehensive evaluation metrics for RAG performance Developed an automated system to test multiple RAG configurations Built a dashboard for visualizing evaluation results and comparisons Integrated the system with cloud services for data storage and processing Implemented auto-deployment of the best-performing RAG pipeline to a chatbot interface SOLUTIONS & STRATEGIES   RESULTS AND IMPACTS – ARCHITECTUAL PATTERN  

Table of Contents Toggle OVERVIEWSCOPE OF WORKAI SolutionSOLUTIONS & STRATEGIESRESULTS AND IMPACTS – ARCHITECTUAL PATTERN OVERVIEW Our client, a global tech talent platform connecting Korean developers with international opportunities, needed an intelligent system to standardize and localize professional profiles across markets. The goal was to develop an AI-powered solution that could bridge the gap between Korean and Western resume standards while preserving technical accuracy and cultural nuances. Country: Korea Domain: Professional Career Services AI expertise: LLM, Document Analysis SCOPE OF WORK The project encompasses development of a comprehensive template standardization system with cultural adaptation and ATS optimization. Implementation of an LLM-powered content generation pipeline with translation capabilities and quality controls. Development of a integrated document generation system handling multiple output formats and quality verification. AI Solution The intelligent template manager provides market-specific formats with dynamic cultural adaptation and format optimization. Advanced natural language processor using different LLM (currently used as GPT-4) drives content enhancement with multilingual translation capabilities. High-performance document generator delivers professional outputs across multiple formats with automated quality assurance. SOLUTIONS & STRATEGIES   RESULTS AND IMPACTS – ARCHITECTUAL PATTERN API response: <1s Concurrent generation: 10 resumes System uptime: 99.9% Translation accuracy: 92% Generation time: avg 3 minutes Daily throughput: 20-35 resumes