Daehyun Kim

Summary

I work on CUBRID database engine features across storage, SQL semantics, vector search, and performance. My recent work includes the OOS epic for out-of-line large-column storage and vector type/operator/function support, with side projects in CLI tooling and AI-assisted development workflows.

            B.Eng.
            Computer Science and Engineering, Korea University
          
            B.A.
            Economics, McGill University; minor in Finance
          
            Focus
            CUBRID engine development, storage internals, vector search
          
            Builds
            OOS storage, vector SQL support, profiling-driven performance fixes

Selected Work

CUBRID Out-of-line Overflow Storage

CBRD-26357 · CUBRID database engine

Driving the OOS epic to store large column values out of line instead of always keeping every column inline in heap records, reducing unnecessary I/O and improving heap page density for projection-heavy scans.

Designed around heap records holding compact OOS references while large values live in a separate OOS area.
Worked through MVCC, recovery, HA, and existing overflow-record consistency constraints.
Maintained CUBRID OOS Project Knowledge Base and CUBRID Code Analysis notes to make the design reviewable.

CUBRID Vector Type, Operators, and ANN Performance

CUBVEC · vector search

Implemented vector type support and vector-related SQL operators/functions in CUBRID, then worked on ANN benchmark performance as vector dimensionality increased.

Added database-facing vector semantics: type handling plus vector operators/functions for SQL use.
Investigated CUBVEC-99 performance gaps between 20-dimensional and 256-dimensional ANN queries.
Applied SIMD/vectorization build options to the usearch distance metric path in CUBVEC-108.

CUBRID Development Tooling

Side projects · Rust / Shell / AI automation

Built supporting tools around day-to-day CUBRID development: code navigation, test triage, JIRA/CI automation, OOS knowledge search, and operational utilities.

patchers

Side project · Rust / terminal UI

Terminal-based interactive tool for reviewing and slicing large diff/patch files into smaller, reviewable hunks for cleaner commits and patch sharing.

Markdown to Jira/Confluence

Side project · Rust / WebAssembly

Browser-based utility for converting Markdown into Jira/Confluence-friendly markup, aimed at reducing friction when moving technical notes into team documentation systems.

Education

Korea University

2019.03 – 2023.02

B.Eng. in Computer Science and Engineering · GPA 4.03 / 4.5

Coursework: operating systems, algorithms, computer architecture, databases, networks.
Programming languages coursework included an interpreter, type checker, and garbage collector.

McGill University

2014.09 – 2018.08

B.A. in Economics · Minor in Finance · GPA 3.11 / 4.0

Completed CS coursework in software systems, computer science, and foundations of programming.
Thesis: Skewness in Individual U.S. Stock Log Returns at Different Investment Horizons.

Naver Connect Foundation Boostcamp

2023.06 – 2023.12

Web/Mobile 8th cohort

EIB The Victor Hugo School, Paris

2011.09 – 2014.06

International Baccalaureate Diploma

Interests

Database systems Storage engines MVCC and concurrency Vector search Runtime systems Programming languages Developer tooling AI-assisted software engineering

Skills

Primary

Rust, C++, C, SQL

Systems

Linux, POSIX, gdb, perf, uftrace, strace, valgrind, make, cmake, docker

Tooling

nushell, git, GitHub Actions, bash, zsh, Neovim, fzf, rg, jq, direnv, just

Platforms

Arch Linux, Rocky Linux, Ubuntu

Familiar

Python, Go, TypeScript, Java, OCaml, Node.js, MySQL

Contact

Open to software engineering roles, research engineering roles, and graduate programs where systems, databases, or programming languages are central.

kimdhyungg@gmail.com linkedin.com/in/daehyun-kim-519419107 github.com/vimkim velog.io/@dkvlg

Download CV PDF: A4 Letter