Youngho Seo

I’m an NLP and applied AI engineer with 5+ years of experience building research-backed educational technology systems across NSF and IES projects. I focus on turning complex research requirements into reliable pipelines that work at scale.

In my current role, I’ve led automatic English item generation—covering vocabulary, phonological, morphological, and sentence-relation assessment items—and supported end-to-end deployment with QA and psychometric analyses including IRT. I also analyze large-scale learning data, processing millions of responses and building analytics and visualization to inform K-12 scoring, timing, difficulty thresholds, and recommendations.

I’ve driven model development through a $1M NSF SBIR effort to build a Transformer-based English text simplification system in Python, achieving near-state-of-the-art results and expanding evaluation coverage while maintaining strong lexical precision. I’ve also built predictive proficiency models that reached up to 80% accuracy for data-driven intervention planning.

Beyond education, I ship product-grade, privacy-focused DICOM tooling: a local-first cross-platform medical imaging viewer for iOS, macOS, Android, and web with on-device file handling and no server upload. I’ve implemented performance optimizations for large DICOM folders and delivered multilingual UI and release-ready assets across App Store and Google Play.