Computational Linguist · Spanish Instructor · Ph.D. Candidate, Digital Learning & Education, NTUST
Applied linguist who builds NLP systems and ships them as working software, not just papers: a hybrid readability model (238 linguistic features + transformer embeddings, ~87% CEFR classification accuracy), a mobile health app, and a self-founded EdTech platform with real users. A decade of classroom experience across three countries is the domain expertise the models are built on — not a separate career.
Conceived, built, and run a digital extensive-reading platform for Spanish learners: promotional landing pages, registration and onboarding flow, and the content pipeline, on a Flask stack I built and maintain myself. Also served as EdTech Project Contributor for The Other Way Spanish, extending the same model into a community reading club.
Dissertation: Assessing Text Readability in Spanish as a Foreign Language: Computational Modelling and Human Judgment. Built a hybrid readability model (linguistic features + transformer embeddings) and validated it against real learner perception, including affective variables like anxiety and engagement.
Model outperforms MultiAzterTest (existing benchmark) on CEFR discrimination while correlating strongly with it (ρ = 0.738). Presented at Edge Hill University (UK), Teachers College Columbia University, Westminster International University in Tashkent, and National Taipei University of Business.
Thesis compared graded readers vs. authentic illustrated texts as first material for beginning Spanish learners: a corpus analysis of verb variety, frequency, mood, and tense, plus a 12-week study with Taiwanese Spanish majors. 170-page manuscript; advisor Dra. Sy-Ying Lee. Now under revision for the journal System (co-authored).
The hybrid architecture behind the dissertation's Study 1 — traditional linguistic features fused with transformer embeddings, trained on a cleaned custom corpus of 800+ Spanish texts. Deployed for public use.
mHealth app for people navigating rare or hard-to-diagnose conditions — structured symptom tracking and a timeline patients can hand to a new specialist, with language built around how patients actually describe symptoms. Currently gathering data toward a published case study.
Reading tool for learners of Chinese that colors characters to make components, radicals, or pronunciation patterns visible at a glance — same readability instinct applied to a language she's still learning herself.
A data model for handling diverse real-world name structures in software — built to acknowledge naming conventions most systems still get wrong.
18 teaching roles across Chile, the U.S., and Taiwan (2015–present): K-12, adult education, DELE exam prep, corporate language training, and university-level EMI support.
| 2023– | English Teacher, Gram Cram School — grades 1–6 (New Taipei) |
| 2024–26 | Teaching Assistant, Applied Foreign Languages, NTUST — helped bring ChatGPT into a Reading & Writing classroom |
| 2021–24 | Spanish & English Teacher, Lion and Lion — DELE B1/B2 exam prep (Taipei) |
| 2021 | EMI Teaching Assistant, NTUST — won Best Tutor award |
| 2019–21 | Private Spanish Teacher, freelance (Taipei) |
| 2018–19 | Spanish Teacher, CEE El Golf — corporate training for Japanese mining-sector learners (Santiago) |
| 2017–18 | Spanish Language Assistant, Union College (Schenectady, NY) |
| 2015–19 | K-12 and adult Spanish/DELE instruction across 9 additional institutions in Santiago, Chile |
Condensed — full 18-role teaching history on the full site.
Co-authored with Sy-Ying Lee. In active revision for the journal System.
Valenzuela & Toledo (2019). Logos: Revista de Lingüística, Filosofía y Literatura, 29(2), 268–285.
Reviewer, Revista Literatura y Lingüística (since 2022). Two further papers on the dissertation studies are queued for the XII Congreso Internacional de la AAH.