# Kai Park: full CV > Full text of the public CV of Kai (Hyeyeon) Park, computational linguist working on language, cognition and LLM alignment. Plain-text mirror of https://bakeyeon.github.io/cv/Kai_Park_CV_public.pdf (updated November 2026). Short overview: https://bakeyeon.github.io/llms.txt - Name: Kai Park (legal name: Hyeyeon Park) - Based: Germany, open to relocation - Work authorization: Germany (post-study residence permit), no visa sponsorship needed - Availability: from January 2027, for full-time research roles in industry or academia, and for PhD positions - Email: kai.hyeyeon.park@gmail.com - Website: https://bakeyeon.github.io/ - LinkedIn: https://www.linkedin.com/in/bakeyeon/ - Google Scholar: https://scholar.google.com/citations?user=CzCE0C0AAAAJ - ORCID: https://orcid.org/0000-0002-2854-0127 - GitHub: https://github.com/bakeyeon ## Professional summary Computational linguist working at the intersection of empirical linguistics and language model evaluation. Studies how linguistic structure shapes cognition and affect, using corpus-based quantitative analysis and large controlled experiments, and turns pragmatic, emotional and cultural nuance into measurable signals that show where model language aligns with human language and where it diverges. Owns research end to end, from experimental design and platform engineering to statistics and sole-authored publication (PANDORA @ EMNLP 2026). Research on modal particles and emoji in translation was featured by DeepL in its DeepL Bridges launch materials, alongside hands-on RLHF data work for Korean LLM alignment. Works across Korean, English and German, with the longer aim of alignment methods that hold across languages, not just within one. ## Core competencies - AI and product strategy: language model alignment, RLHF annotation, human-in-the-loop evaluation, benchmark design, cross-functional stakeholder management - Tools and frameworks: statistical modeling (R/Python), cloud resource orchestration (GCP), empirical experimental design - Languages: Korean (native), English (C1 / professional), German (C1 / professional) ## Publications and key research - Park, K. (2026). The Grue Divide: When Cognitive Accuracy Doesn't Predict Affective Alignment. Poster, PANDORA @ EMNLP 2026. Sole author; developed from the author's MA thesis. Paper: https://aclanthology.org/2026.pandora-1.2 · Experiment platform: https://bakeyeon.github.io/colorfish/ - MA thesis: "Can the Use of Derivatives Affect Cognition? A Large-Scale Empirical Study on Morphological Derivation and the Korean Blue-Green Spectrum." University of Göttingen, 2026. https://publications.goettingen-research-online.de/handle/2/182236 - Park, K. (2024). Modalpartikeln: Can Sentiments Survive Translation with Emojis? Term paper, HHU Düsseldorf. https://doi.org/10.5281/zenodo.24326143 - Park, K. (2025). How Much Sentiment Can Be Carried by Modal Particles? Term paper, HHU Düsseldorf. https://doi.org/10.5281/zenodo.24326464 - Public-facing research writing: "Can We Translate Our Sentiments?" (GoPenAI on Medium, 2026). An accessible piece revisiting 2024 term-paper research on cross-lingual emotional nuance and perceptual color cognition, featured in DeepL Bridges launch materials (https://www.deepl.com/en/blog/connecting-insights-deepl-bridges). https://blog.gopenai.com/can-we-translate-our-sentiments-140c55822ead ## Research and product framework projects ### Large-scale cognitive data architecture (MA thesis project and PANDORA paper) - Designed, built and deployed a web-based empirical experiment platform, collecting 9,534 analyzed trials from 1,159 participants (about 2,200 recruited); formulated statistical models and data pipelines to quantify morphological cognition effects. ### Light Verbs, Heavy Meaning? (MA seminar term paper, Göttingen) - Tested Polinsky and Magyar's (2020) informativity account of head-final lexicons on Korean simplex verbs vs. root+hata constructions via dictionary sense counts, embedding self-similarity (KLUE-BERT, KcBERT, Vertex AI) and corpus frequency. The predicted asymmetry held in 58% of 38 conditions and was item-driven rather than structurally deterministic. https://github.com/bakeyeon/light-verbs-heavy-meaning ### Modal particles, emojis and sentiment translation (MA term paper series, HHU Düsseldorf) - Survey study finding a significant correlation between the emotional scores of German modal particles and expressive emojis, extended with a benchmark of FastText, Hugging Face and BERT sentiment models. https://github.com/bakeyeon/Modalparticles_Emojis · https://github.com/bakeyeon/modalparticles_sentiment_analysis - Built "Sentiment Translator", a Gemini/AI Studio prototype visualizing cross-lingual sentiment shifts via an interactive emoji-based emotion map. https://github.com/bakeyeon/sentiment_translator ## Work experience ### Freelance Linguistic Editor, AI Alignment (Korean/English): DeepL SE, Language Data Team, Research (Nov 2024 – present, remote) - Human-in-the-loop RLHF alignment: preference ranking, minimal-error editing and full rewriting to build gold-standard datasets for Korean LLM alignment. Formulated structural linguistic rationales and proposed new decision criteria for cross-lingual semantic ambiguity (KO–EN). - Part of DeepL Bridges Early Access; research on linguistic nuance featured in official platform launch materials, including work on modal-particle sentiment and emoji-mediated translation, and the survey platform built for the "Grue Divide" paper (poster, PANDORA workshop, EMNLP 2026). ### Freelance Localization QA Specialist and Translator: CCJK Technologies (Marstranslation) / SUCCESSGLO PTE LTD (May 2022 – Nov 2026, remote) - Delivered linguistic QA, translation, voice-over (tech/legal/compliance) and full/technical MTPE across Korean–English and German–Korean pairs; served as final QA on 100+ tasks, including Korean voiceover localization for English IVR systems. - Handled UI-based length-overflow issues in German–English–Korean localization for automotive infotainment clients, and provided structured peer review for junior linguists, working with Trados and proprietary QA tools. ### Head of Strategic Planning and Brand Strategist: Minor to Winner, social venture (May 2016 – Jul 2016, on-site) - Led strategic planning as one of two core team members alongside the founder during early-stage company building, with cross-functional ownership of operations, branding and HR. - Scaled the team by onboarding 8 new members within the 3-month mandate. - Co-led rebranding initiatives for regional industry promotion. - Secured government grant funding and led contract negotiations with public and private stakeholders. ### Performance Marketing Planner (AE), internship: 9 Fruits Media (Dec 2015 – May 2016, on-site) - Owned budget planning and allocation for multi-channel growth campaigns on a major TV home-shopping platform, directing creative-team briefs and deliverables end to end. - Drove measurable conversion growth via data-driven content optimization and cross-platform tracking: Facebook purchase-link clicks +173% in 2 months (829 → 2,265); Kakao conversion clicks +516% in 6 months (25 → 154); Naver link clicks +408% in 6 months (123 → 625). - Managed publication workflows and authored monthly performance and competitor-analysis reports for client stakeholders. - Led client-facing communication and marketing analytics. ## Education - M.A. in Linguistics, Georg-August-Universität Göttingen, 2023–2026. Most coursework completed at Heinrich Heine University Düsseldorf; degree conferred by Göttingen. GPA 1.4 (German scale, 1.0 best; "sehr gut"); MA thesis grade 1.0. - Pre-MBA (Social Entrepreneurship Management Program), KAIST College of Business, 2016. Selected as key talent alongside the CEO of Minor to Winner; program fully sponsored by SK Group. - B.A. in English Literature and Languages (minor: Mass Communication), Pusan National University, 2011–2015, including a winter exchange term at Seoul National University (2013). - Teaching Korean as a Foreign Language Program, NILE Korea, 2023–2024. - Coursework in Computer Science (Informatik), TU Darmstadt, 2020–2023. ## Honors and awards - Google Cloud TPU Research Cloud grantee (2026). Architected and open-sourced a Python-based Spot TPU orchestration tool that streamlines environment setup for fellow TRC researchers. https://github.com/bakeyeon/gcloud-tpu-spot-automator - The Next Century Humanities Scholarship: full-ride national scholarship, Ministry of Education, Korea (2011). - Invited to the Frontier Research Club 2026 series. - Cannes Lions Korea and Innocean Worldwide, Busan Region Award (2014): team lead and main presenter for a 3-person team; conceived and presented an integrated PR campaign on teen smartphone overuse. - Creative Arena, 3rd place, national final (2014): team lead and main presenter for a 3-person team; presented an AMD brand-reputation campaign live to 5,000+ attendees (COEX, broadcast on AfreecaTV). A proposal concept from the competition was later implemented on the organizer Stu:m Creation's official platform. - KT&G Marketing Competition, National Excellence Award (2013): team leader and lead planner; won multiple first-place regional rounds on the way to the national award. - Anti-Corruption and Civil Rights Commission of Korea, Top Prize for Online PR (2012): owned monthly budget allocation for a multi-institution PR campaign; grew a Naver blog to 1M+ page views and led a 1,000-attendee offline campaign. ## Certifications - Teaching Korean as a Foreign Language (Level 2), National Institute of Korean Language (2024) - Language certifications: EF SET English (C1 Advanced, 2024) · TestDaF (CEFR C1, 2019) · IELTS Academic (CEFR C1, 2023) ## Talks - Invited speaker, APAC AI for Science DevFest 2026 (Google Researcher Groups, GDG AI for Science), Nov 27–28, 2026. https://bakeyeon.github.io/talk.html