Structured Data Structured Data Design
Schema and relational design for structured inputs including EMR and feature tables.
Portfolio for Technical Review
Data & Applied AI Engineer
Projects and research records across data structure, AI development workflows, and NLP/LLM evaluation.
Each project is framed by its problem, design decision, implementation result, and available artifacts.
Review Mode
Choose a first reading angle
Start with official record and role fit, then use cases for technical support.
Focus Areas
Reorder representative work by focus area
Choose data, AI workflow, or NLP/LLM to bring the most relevant projects forward.
Choose a focus area to bring relevant projects forward.
Development Flow
Recurring development flow across AI projects
This summarizes recurring work patterns across projects, not a click-by-click stage diagram for one project.
Text & Embedding Text & Embedding
Tokenization and semantic embedding space structure for unstructured text.
Knowledge Graph Knowledge Graph
Real-time knowledge graph design compatible with GNNs and recommender systems.
Hybrid Pipeline Hybrid Pipeline
Integrating heterogeneous sources into high-performance processing pipelines.
Data Data surface
Shape structured, text, image, and graph inputs into usable modeling surfaces.
Experiment Tracking
Keep hypotheses, settings, metrics, and artifacts inspectable.
Training Training flow
Manage model training through reproducible environments and workflows.
Evaluation Evaluation design
Review errors, rare cases, and domain needs beyond top-line scores.
Deployment Delivery
Package outputs into reviewable artifacts and execution surfaces.
Observability Visibility
Use logs, artifacts, and run records to explain system state.
Feedback / Recovery Recovery
Connect comparison, recovery, and retrospection into the next cycle.
Language Modeling Language Modeling
Domain-specific pre-training and downstream task adaptation/fine-tuning.
Evaluation Design Evaluation Design
Multi-dimensional evaluation addressing rare errors, biases, and domain constraints beyond simple metrics.
Alignment & Preference Alignment & Preference
Instruction tuning and alignment with human/AI preferences via RLAIF/RLHF.
Research-to-System Research-to-System
Translating advanced NLP research into production-grade web service APIs and lightweight optimized models.
Project Records
Project records
Project title and explanation stay separate from metadata such as data surfaces and availability.
Data Systems
Data Systems Specialist with Graph-native Depth
Turns heterogeneous data into model-ready and system-ready structures, pipelines, and graph representations.
Contexta: Local-First ML Observability
Shows operational observability design for tracing, comparing, and recovering AI execution records and artifacts.
Lynxes: Graph Analytics Engine
Shows system design and implementation depth for treating graph data as a first-class execution model.
EMR-Based Nursing Surveillance for Automatic ICD Coding
Combines heterogeneous structured EMR and Korean clinical text into an evaluable NLP pipeline.
Dalkom Shop: Internal Employee Mileage Commerce Platform
Shows delivery, security, and observability foundations for running service features in an operational environment.
AI-DLC / MLOps
AI-DLC and Operational MLOps Engineer
Makes run records, artifacts, model behavior, and feedback paths inspectable enough to improve.
Contexta: Local-First ML Observability
Shows operational observability design for tracing, comparing, and recovering AI execution records and artifacts.
Lynxes: Graph Analytics Engine
Shows system design and implementation depth for treating graph data as a first-class execution model.
Dalkom Shop: Internal Employee Mileage Commerce Platform
Shows delivery, security, and observability foundations for running service features in an operational environment.
Devridge: LLM-Based Feedback Bridge for Developers
Shows prompt and interaction design for making LLM output useful in role-based technical review.
BloGeek: AI Modules for a React + Spring Blog Project
Connects Korean NLP classification and generation models to product-shaped web service features.
FRIMO: Conversational AI for Emotional Support and Diary Generation
Connects Korean NLP models into a user-facing AI pipeline for conversational product experience.
NLP / LLM
Applied NLP and LLM Research Engineer
Translates language-model research into better modeling and evaluation decisions in applied systems.
EMR-Based Nursing Surveillance for Automatic ICD Coding
Combines heterogeneous structured EMR and Korean clinical text into an evaluable NLP pipeline.
Devridge: LLM-Based Feedback Bridge for Developers
Shows prompt and interaction design for making LLM output useful in role-based technical review.
BloGeek: AI Modules for a React + Spring Blog Project
Connects Korean NLP classification and generation models to product-shaped web service features.
FRIMO: Conversational AI for Emotional Support and Diary Generation
Connects Korean NLP models into a user-facing AI pipeline for conversational product experience.
Project Comparison
Featured work framed by problem, decision, and outcome
The portfolio is not a project list; it is a map of judgment, tradeoffs, and results.
Project Comparison
Compare representative projects by the same criteria
Review featured work by problem, decision, outcome, and availability before reading the full case.
| Case | Problem | Decision | Outcome | Availability | Open |
|---|---|---|---|---|---|
| Lynxes: Graph Analytics Engine Graph Systems Engine Project / 2026 | Existing Python graph libraries and generic dataframe wrappers often struggle to combine memory efficiency, traversal performance, and lazy query optimization for large graph analytics. | Designed GraphFrame to own Arrow RecordBatches directly. | Established the foundation for a graph analytics engine with Arrow columnar memory, CSR-based traversal, and lazy collect execution. | Built project Public | Open |
| Contexta: Local-First ML Observability Self-directed ML Platform Project / 2026 | ML experiments and deployment work often scatter metadata, records, and artifacts across tools, making reproducible local observability hard to maintain. | Used a `.contexta/` workspace as the home for separated metadata, records, and artifact storage. | Implemented a local observability foundation for consistently managing and inspecting ML execution history and artifacts. | Built project Public | Open |
| BloGeek: AI Modules for a React + Spring Blog Project Collaborative NLP Project / 2023 | The product needed ML components that could classify emotional polarity and generate stylistic variations of text to support richer blog content workflows. | Used KoBERT for polarity recognition and KoBART for style transfer. | The project gave the team practical AI modules for blog-oriented text processing. | Built project Summary available | Open |
Graph Systems Engine Project / 2026
Lynxes: Graph Analytics Engine
A high-performance graph analytics engine that combines Arrow columnar memory with graph-native traversal structures for Python users.
Existing Python graph libraries and generic dataframe wrappers often struggle to combine memory efficiency, traversal performance, and lazy query optimization for large graph analytics.
Designed GraphFrame to own Arrow RecordBatches directly.
Established the foundation for a graph analytics engine with Arrow columnar memory, CSR-based traversal, and lazy collect execution.
Self-directed ML Platform Project / 2026
Contexta: Local-First ML Observability
A local-first observability library for tracing, comparing, and recovering ML execution history through one consistent contract.
ML experiments and deployment work often scatter metadata, records, and artifacts across tools, making reproducible local observability hard to maintain.
Used a `.contexta/` workspace as the home for separated metadata, records, and artifact storage.
Implemented a local observability foundation for consistently managing and inspecting ML execution history and artifacts.
Collaborative NLP Project / 2023
BloGeek: AI Modules for a React + Spring Blog Project
A Korean NLP project connecting emotion classification and style transfer models to a blog product workflow.
The product needed ML components that could classify emotional polarity and generate stylistic variations of text to support richer blog content workflows.
Used KoBERT for polarity recognition and KoBART for style transfer.
The project gave the team practical AI modules for blog-oriented text processing.
Process
A repeatable way of turning research into systems
This section shows how research, experimentation, implementation, and delivery connect.
Feature Engineering
Data structure and management
I preprocess complex data in ways that fit the task and design pipelines that preserve data quality and consistency.
Reproducible Experiments
Automating training
I train models under carefully controlled code and environment settings, and track them systematically to build experiments that can be reproduced at any time.
Robust Evaluation
Validation and assessment
Going beyond simple accuracy, I examine robustness, error cases, and application-context requirements from multiple angles before practical use.
DevOps
Infrastructure and deployment
I deploy models in AWS or Docker environments and support reliable operation in real-world settings through continuous integration and automation.
System Observability
Monitoring and feedback loops
I collect and visualize logs, resource signals, and prediction outputs from deployed AI systems in real time so that the internal state of the pipeline can be observed transparently.
Research
Research records
Research is connected as modeling and evaluation background behind project decisions.
Deep Learning based Automatic ICD Coding for Nursing Surveillance of Abdominal Surgery Patients
Supports the portfolio claim that NLP/LLM systems should be judged through domain data structure and error distribution, not only headline accuracy.
EMR-Based Nursing Surveillance for Automatic ICD Coding
Recognition
Learning and practice signals behind the work
Certificates and awards support the work record rather than replacing it.
Practical Implementation of Monitoring and Testing in DevOps Environments
LLOYDK / 2023.11
Completed practical training in Elastic-based DevOps monitoring and testing.
Multi Cloud Orchestration Program
5Works / 2023.12
Completed HashiCorp-based multi-cloud orchestration and IaC training.
Company-Led Intensive Project Training
DK Techin / 2024.02
Participated in industry-linked practical training focused on security and DevOps engineering.
Micro Degree in Software Specialist Training
Gachon University / 2024.02
Completed a micro-degree program for training software specialists.
Use the resume for formal review and the cases for technical depth
I work across data structure, AI development workflows, and NLP/LLM evaluation, carrying problems from framing to implementation, validation, and delivery.