Jerome

As a 5+ year full-stack software engineer, my strengths lie in architectural improvement and performance optimization of large-scale IoT control systems, building AI-driven services, and translating complex real-world domains into working code through AI.

[I enjoy finding the best architecture within given constraints.] Since starting my career at a startup, I have never accepted limited headcount or resources as an inevitable ceiling — instead, I dig deep to explore and optimize the best possible structure within those bounds. I redesigned an IoT lighting control system that topped out at 10,000 devices per single Pod to handle over 1,000,000 devices, and built a RAG-based AI agent that cut the business team's email response time by 95%.

[I weigh the trade-offs between user experience and technical value before deciding.] When choosing and adopting technology, I always consider user experience, practical utility, and cost minimization together. Matching the nature of the documents powering AI responses, I chose an in-memory Vector Store over an external Vector DB to eliminate unnecessary infrastructure costs — and replaced an external paid monitoring solution with an in-house dashboard, cutting ₩6M in monthly operating expenses.

[I proactively surface problems and pursue smarter ways of working.] Starting from the essential question — "why is this needed?" — I analyze work from multiple angles, surface potential issues early, and lead improvements. A representative example: I identified the security risks of directly linking employee personal Gmail accounts and drove the adoption of a company-wide shared account and proxy server. Recently, I have been focused on raising engineering productivity and product quality through AI-native development — applying Context Engineering to AI Agents like Claude Code.