TL;DR I build AI agents and supporting systems at Upstage.
Read this one-pager to learn about my work and interests. For more detail, see my CV or contact me for a chat on LinkedIn.
I’m an AI engineer on Upstage’s Product Engine team. We build production-ready AI agents and supporting systems — robust to non-determinism while fully leveraging LLMs’ generative capacity.
Currently, I’m building agents that help users with document-centric tasks in our AI Space.
Previously, I automated qualitative evaluations for our LLM, Solar, to ensure its behavior aligned with our goals. I also applied RAG to build a chat engine for the virtual K-pop idol MAVE:, serving thousands of users worldwide.
Before Upstage, I was an AI researcher at Silvia Health, training models to detect cognitive impairment from smartphone speech.
I completed a master’s degree in the Interactive Computing Lab at KAIST. I explored active learning for data-efficient emotion recognition (AdaptiveESM) and created the K-EmoCon Dataset, a multimodal, in-the-wild biometric corpus for emotion recognition.
I also dabble in Web3 — trading, building, and exploring DeFi. Earlier this year, I created alignment-protocol, an incentive-alignment system for AI data, for the Solana Colosseum hackathon. The protocol is at ArVxFdoxzCsMDb1K3jXsQTrDP4mbfHMxKiZLjZpznB5c on the Solana devnet.
When I’m away from the laptop, you’ll usually find me at the climbing gym. I recently started practicing Vipassana meditation.
Last updated: 2025-10-17