Role
AI Engineer
EUISUH JEONG
aka
[uiseo]
Engineer by training, researcher by drift.
A short bio on origins, trajectory, and the cultures that shaped the way I read systems.
bornSeoul, KR
countries4
fluentKO · EN
workingHI · AR (some)
stackPython · Torch
edgeFastAPI · Celery
infraRedis · Docker
I'm an AI engineer and software engineer currently serving as a Staff Sergeant with the Republic of Korea Air Force, where I build computer-vision systems for runway integrity.
Before the service, I helped found aiXamine at QCRI — a platform that stress-tests language models against safety benchmarks. I'm a Carnegie Mellon CS '22 grad, with a minor in Mathematical Sciences.
I've been moving since I was three. Seoul, then a small town in the US, then back to Seoul, then India for secondary, then CMU, then Qatar for work, then home again. Cultures stack, like middleware. The interesting work happens in the seams.
What I'm doing this week.
Updated automatically.
Two papers out. Two manuscripts in progress.
The ROKAF runway work produced two KOSAP papers in 2025. I am now preparing TraceGuardBench, on tool-use safety in coding agents, and RepOrbit, on interpretable repository-management analysis. Military service wraps up in Q4.
- Published: ROKAF Runway Crack Dataset — KOSAP Vol. 1 No. 2 (Dec 2025)
- Published: Deep Learning for PCI Assessment — KOSAP Vol. 1 No. 1 (Aug 2025)
- In preparation: How Far Do Lightweight Pattern Guards Get for Coding-Agent Tool Safety?
- In preparation: RepOrbit: An Interpretable Event-Sourced Index for Repository Management
- Pulling Korean military service to a close in Q4 — looking for what comes next
Six years, three time zones.
Full record on LinkedIn. Highlights below.
2025 — present
AI Engineer · Staff Sergeant · Squad Leader
Republic of Korea Air Force / AI-Based Technology Team
Led the squad that built and deployed an AI-driven runway pavement evaluation system at an active ROKAF airbase. Constructed a 231,347-image dataset — 52,800 real captures augmented with 178,547 alpha-blended synthetic images across 9 defect classes (SSIM 0.98163, FID 4.2145) — published as the ROKAF Runway Crack Dataset in KOSAP (Vol. 1, No. 2, Dec 2025) as first author. Co-designed the PCI scoring pipeline around YOLOv11, achieving 86.8% detection accuracy and a 98.2% reduction in manual assessment time; published in KOSAP (Vol. 1, No. 1, Aug 2025). Full project lifecycle as technical lead.
2023 — 2025
Research Engineer
Qatar Computing Research Institute (QCRI)
Co-developed aiXamine — a black-box LLM safety evaluation platform with 40+ benchmarks across 8 security dimensions. Built the modular reporting + visualization architecture; evaluated 50+ models across 2K+ exams, surfacing vulnerabilities in GPT-4o, Grok-3, and Gemini 2.0. Also investigated backdoor Trojan attacks on code-focused LLMs (finetuning + susceptibility testing).
2022 — 2023
Software Engineer
KARTY · Spend, Save, and Manage
Built a multi-channel notification system (SMS, email, push) for the consumer fintech app. Migrated payment processing to a compliant platform under regulatory scrutiny. Designed and shipped a Clubhouse-style waiting list + lottery system tied to FIFA World Cup Qatar 2022.
2021 — 2022
Teaching Assistant · 11-785 Deep Learning (PhD-level)
Carnegie Mellon University · Pittsburgh
Planned and delivered lectures, recitations, and assignments to 350+ students in CMU's flagship deep-learning course. Mentored research projects and guided exploration of novel directions. Sample recitation on YouTube →
2018 — 2022
B.S. Computer Science · Minor, Mathematical Sciences · University Honors
Carnegie Mellon University
Coursework concentrated in systems, machine learning, and applied math.
Things I built that went live.
A non-exhaustive list. The ones I can talk about are below; ask for the rest.
Runway Evaluation System
Live · ROKAFTech lead · data + model + backend
Detects cracks and surface defects on airbase runways and computes PCI scores from high-res imagery. In operational use — 86.8% detection accuracy, two KOSAP papers published.
YOLOv11Co-DETRFastAPICeleryRedisPostgresDocker
aiXamine
Live · publicFounding member · benchmark harness
A black-box LLM safety platform that runs repeatable exams across bias, robustness, jailbreaks, and other risk dimensions. I helped build the benchmark harness and reporting system.
PythonLLMEvalBenchmarks
Papers and conference work.
Manuscripts in preparation, peer-reviewed publications, and conference work · newest first.
Manuscript in preparation · 2026
How Far Do Lightweight Pattern Guards Get for Coding-Agent Tool Safety?
E. Jeong
TraceGuardBench measures unsafe tool calls in coding-agent traces and tests lightweight, task-scoped guards before execution. It covers secret access, destructive shell use, network exfiltration, and instructions embedded in untrusted tool output.
Manuscript in preparation · 2026
RepOrbit: An Interpretable Event-Sourced Index for Repository Management
E. Jeong
RepOrbit turns Git and GitHub history into reproducible repository timelapses and explainable Repository Management Quality Index (RMI) trajectories, connecting code changes with issues, reviews, CI, releases, and maintenance signals.
KSMI · 2026 · Extended abstract
Music Tagging Graph Neural Network with Tag Labels
Y. Park · J. Park · E. Jeong
MTGNN is a graph neural network framework for music auto-tagging. It adapts ATGNN's graph-based audio-tagging idea to music by redesigning node generation around semantic and timbre features, then uses a CLAP-initialized Graph Transformer to model dependencies between tag labels.
KOSAP · Vol. 1, No. 2 · Dec 2025
ROKAF Runway Crack Dataset: Construction and Application of a Large-Scale AI-Based Runway Defect Detection Dataset
E. Jeong · S. Ji · M. Kim · H. Lee
A large-scale runway defect dataset built from real airfield captures and synthetic augmentation for AI-based crack and surface-defect detection.
KOSAP · Vol. 1, No. 1 · Aug 2025
Deep Learning for Pavement Management System: Proposing an Automated Pipeline for Pavement Condition Index (PCI) Assessment
S. Ji · E. Jeong · M. Kim · H. Lee
An automated pavement-management pipeline that combines runway defect detection with PCI scoring to reduce manual inspection work.
arXiv · Apr 2025
aiXamine: A Comprehensive Safety Evaluation Platform for Large Language Models
… · E. Jeong · … (see paper for full author list)
A safety-evaluation platform for large language models, covering bias, robustness, jailbreak, and other benchmark-driven risk checks.
Pictures.
A small visual log of places that shaped the work: Seoul, India, Doha, Pittsburgh, and back to Seoul.
1998
SEOUL · born here
2001
USA · kindergarten
2004
SEOUL · elementary
2010
INDIA · secondary
2018
DOHA · CMU-Q
2020
PITTSBURGH · CMU main
2022
DOHA · QCRI · aiXamine
2025
SEOUL · service
Get in touch.
Open to research collaborators, post-service roles, and the occasional good email. Fastest reply on LinkedIn.