Portfolio Document
Applied AI Researcher
Donghyeon Kim
AI Researcher Portfolio
Portfolio / 2026
I build AI that is accurate, understandable, and reusable.
Applied AI Researcher
Donghyeon Kim
Applied AI researcher and AI engineer focused on deployable conversational AI and medical AI systems
I am interested in applied AI problems grounded in real-world data, and in connecting them to models, evaluation, and deliverables that collaborators can actually use. Recently, I have been working across conversational AI, medical AI, and ML systems with a focus on improving reproducibility and practical delivery.
I am a graduate researcher with experience in NLP, medical AI, deep learning, and end-to-end AI project implementation. I am particularly interested in turning experimental AI work into trustworthy systems through stronger evaluation, clear documentation, and a practical MLOps perspective.
Applied AI ResearcherData ScienceApplied AIMLOpsObservability in AI
Working Principles
I build AI that is accurate, understandable, and reusable.
A working principle that guides how research becomes implementation.
I treat evaluation as part of the deliverable, not an appendix.
A working principle that guides how research becomes implementation.
I connect experimentation and delivery beyond individual ownership.
A working principle that guides how research becomes implementation.
Background & Foundations
This portfolio is structured in a reading- and print-friendly format that first presents a short overview, followed by case study sheets for each project.
The selected work is organized around applied AI problems in which modeling, interpretation, and delivery were all important.
Experience.
Intelligent Data Analytics Lab., Gachon University
2024.03 - 2026.02
- Led graduate research on EMR-based nursing surveillance decision support and diagnostic classification.
- Built end-to-end modeling pipelines using KM-BERT ensembles, XGBoost, and both structured and text data.
- Explored conversational AI topics, including empathetic dialogue generation with AI-based feedback.
National Research Foundation of Korea (NRF)
2024.03 - 2025.12
- Contributed to an NRF-funded clinical AI project centered on nursing surveillance decision support using EMR data.
- Implemented workflows for clinical text understanding, including keyword extraction, dependency parsing-based preprocessing, topic modeling, and similarity analysis.
- Provided interpretable results and web-based analysis interfaces for collaborative researchers.
- Achieved 92%+ diagnostic prediction accuracy in a multi-label task by ensembling text and structured data models (KM-BERT, XGBoost).
Institute of Information & Communications Technology Planning & Evaluation (IITP)
2025.09 - 2025.12
- Implemented evaluation-related code in a human-centered multimodal AI project.
- Bridged evaluation requirements with actual code and reviewable deliverables.
dktechin
2024.01 - 2024.02
- Participated in an industry-led intensive program focused on cloud, CI/CD, security, and DevOps practices.
- Contributed to implementation tasks with a focus on security and DevOps roles in project-based training.
Gachon University / Notion Community Program
2019.03 - 2025.02
- Held multiple leadership roles in the official university programming club and served as president in 2022.
- Planned and led study groups on machine learning, big data, financial ML, and GNNs.
- Supported campus learning communities and resource sharing as part of the Notion Campus Leader program (24-2 ~ 25-1).
Education & Learning Foundation
Education.
Gachon University
2024.03 - 2026.02
GPA 4.14/4.5. Intelligent Data Analytics Lab. Advisor - OkRan Jeong. Focused on clinical AI, evaluation, and implementation-oriented research delivery.
Gachon University
2019.03 - 2024.02
Gained leadership and community management experience by participating in an official programming club and leading various study groups, including basic ML, advanced ML, financial ML, and GNN.
Study.
NEKA
A study on implementing Keyword Extraction and NER with KoBERT
SUNAROUM
A study on building buy recommendation models from financial statements and indicators
I-DA
A study on recommendation systems using knowledge graphs and GNNs
MLADS
A study on the fundamentals of data science and machine learning
Capability Summary
Data Science
The ability to refine real-world data for reliable modeling through analysis and preprocessing
Data Preparation · Feature Engineering · Data Wrangling · Data Manipulation
Applied AI
The ability to turn domain-specific NLP into concrete tasks through experimentation and implementation
Conversational AI · Medical AI · Knowledge Graph · Logic-based AI
MLOps
The ability to build repeatable workflows and scale them through deployment and automation
Cloud · Docker · Continuous Training · Feature Store
Observability in AI
An operational perspective for tracing model behavior to improve reproducibility and maintenance
Logging · Monitoring · Artifacts · Provenance
Research
A Context-Adaptive Gated Embedding Framework for Advanced Clinical Decision-Making
2026
This study proposes a hierarchical clinical decision support framework that estimates diagnostic context via partial-label automated ICD coding and reinjects it into irregular ICU time-series forecasting through context-adaptive gating for mechanical ventilation transition prediction. By conditioning temporal interpretation on diagnostic context, the framework substantially improves rare-event detection.
Clinical Decision Support System · Automated ICD Coding · ICU Time-series · Mechanical Ventilation Prediction · Partial-Label Learning · Extreme Multi-Class Classification · TCN · Gating · Rare Event Detection
Deep Learning based Automatic ICD Coding for Nursing Surveillance of Abdominal Surgery Patients
2025
This study proposes an automatic ICD coding model for nursing surveillance of abdominal surgery patients by integrating EMR-based test data, patient information, and nursing notes. A stacking architecture combining dual KM-BERT, XGBoost, and PCA outperformed both a single KM-BERT model and simpler ensemble variants.
Medical AI · Nursing Surveillance · EMR · Automatic ICD Coding · Deep Learning · KM-BERT · XGBoost · Ensemble · Abdominal Surgery
Empathetic Dialogue Generation Model Using Reinforcement Learning with AI-Based Feedback
2025
This study proposes an empathetic dialogue generation model using reinforcement learning with AI-based feedback (RLAIF) to address limited diversity and reliance on human feedback. By leveraging an LLM as a reward evaluator and integrating it into EmpRL, the model generates more diverse empathetic responses.
Empathetic Dialogue · Reinforcement Learning · RLAIF · RLHF Alternative · LLM · Dialogue Generation · NLP · AI Feedback
Completion and Awards
Certificates & Qualifications
Practical Implementation of Monitoring and Testing in DevOps Environments
2023.11
Completed practical training in Elastic-based DevOps monitoring and testing.
- Studied DevOps-related services and operational environments provided by Elastic.
- Learned DevOps concepts applicable to practical work from monitoring and testing perspectives.
Certificates & Qualifications
Multi Cloud Orchestration Program
2023.12
Completed HashiCorp-based multi-cloud orchestration and IaC training.
- Studied mainly HashiCorp IaC tools such as Packer and Terraform.
- Learned the concepts of infrastructure automation and orchestration in multi-cloud environments.
Certificates & Qualifications
Company-Led Intensive Project Training
2024.02
Participated in industry-linked practical training focused on security and DevOps engineering.
- Conducted practical projects while working on cloud, CI/CD, and security-related tasks.
- Took on the role of advisor and build assistant as an experienced member during the recommender system development process.
Certificates & Qualifications
Micro Degree in Software Specialist Training
2024.02
Completed a micro-degree program for training software specialists.
- Completed a program aimed at strengthening software and practice-oriented digital capabilities.
- Built a record of completion that connects undergraduate-based education with practice-oriented project experience.
Project Archive
Case Study 1
EMR-Based Nursing Surveillance for Automatic ICD Coding
Built an automatic ICD coding pipeline for nursing surveillance of abdominal surgery patients using core EMR data.
Graduate Researcher · AI Engineer · Data Scientist
Case Study 2
Dalkom Shop: Internal Employee Mileage Commerce Platform
Served as the DevSecOps lead for an internal employee-only commerce platform, building the delivery, operations, security, and observability foundation behind the service.
DevSecOps Engineer · Infra Engineer · Platform Engineer
Case Study 3
Devridge: LLM-Based Feedback Bridge for Developers
Built an LLM prototype that helps solo developers receive contextualized feedback from different professional roles.
LLM Product Builder · Prototype Engineer
Case Study 4
BloGeek: AI Modules for a React + Spring Blog Project
Contributed to the AI side of a React + Spring blog project, focusing on polarity recognition and style transfer for Korean text.
AI Engineer · NLP Engineer
Case Study 5
FRIMO: Conversational AI for Emotional Support and Diary Generation
Contributed to the machine learning pipeline for a conversational diary app, with a primary focus on emotion recognition and Korean language model integration.
AI Engineer · NLP Engineer
Case Study 1
2025EMR-Based Nursing Surveillance for Automatic ICD Coding
Built an automatic ICD coding pipeline for nursing surveillance of abdominal surgery patients using core EMR data.
Problem
Nursing surveillance required diagnosis-related classification, but key clinical signals were fragmented across heterogeneous EMR sources.
Approach
I integrated structured EMR features and nursing text, then used a dual KM-BERT, PCA, and XGBoost stacking architecture for ICD prediction.
Outcome
The final model achieved 0.9245 accuracy and strong rare-class recall without depending on physician-centered post-hoc documents.
Case Study 2
2024Dalkom Shop: Internal Employee Mileage Commerce Platform
Served as the DevSecOps lead for an internal employee-only commerce platform, building the delivery, operations, security, and observability foundation behind the service.
Problem
The project needed more than frontend and backend implementation; it required cloud infrastructure and platform foundations that could reliably support search, notifications, admin workflows, and ongoing operations.
Approach
I focused on the DevSecOps and infrastructure/platform layer, organizing CI/CD, cloud deployment flow, operational setup, and security-aware service foundations for a React + Spring commerce system.
Outcome
The project demonstrated how infrastructure and platform engineering directly shape the reliability and readiness of a production-style internal commerce service.
Case Study 3
2024Devridge: LLM-Based Feedback Bridge for Developers
Built an LLM prototype that helps solo developers receive contextualized feedback from different professional roles.
Problem
Developers working alone often need UI, performance, or code quality feedback, but they rarely have an easy way to gather role-specific input at the right time.
Approach
I framed the service as a role-aware feedback bridge and focused on role constraints, and contextual input design so the generated responses would stay relevant and scoped.
Outcome
The prototype demonstrated how structured prompting and role-specific constraints could turn LLM output into more useful, contextualized project feedback.
Case Study 4
2023BloGeek: AI Modules for a React + Spring Blog Project
Contributed to the AI side of a React + Spring blog project, focusing on polarity recognition and style transfer for Korean text.
Problem
The product needed ML components that could classify emotional polarity and generate stylistic variations of text to support richer blog content workflows.
Approach
I focused on the machine learning side by implementing a KoBERT-based polarity recognition model and a KoBART-based style transfer pipeline, along with dataset handling and repeated model experiments.
Outcome
The project gave the team practical AIs for blog-oriented text processing and strengthened my experience in integrating classification and generation models into a web product context.
Case Study 5
2023FRIMO: Conversational AI for Emotional Support and Diary Generation
Contributed to the machine learning pipeline for a conversational diary app, with a primary focus on emotion recognition and Korean language model integration.
Problem
The product needed ML components that could recognize user emotion and support a diary-generation workflow from daily conversation logs.
Approach
I focused on the ML side by implementing and refining the emotion recognition and by helping connect chatbot and summarization models into the overall AI workflow.
Outcome
The project delivered an MVP-level conversational diary experience and gave me hands-on experience in integrating Korean NLP models for emotion-aware user interaction.