AI Research Engineer

Singon Kim

I'm interested in Foundation Models, AI Agents, and Human-Centered Intelligence — across pre-training, post-training, and data-centric AI, with a focus on building reliable, scalable systems.

I believe AI should reduce knowledge asymmetry and help people overcome their limitations. My goal is to build intelligent systems that make expertise more accessible, amplify human potential, and contribute to a happier, more empowered society.

SingonKim singon-kim CV Portfolio
Singon Kim
01

Experience

Mar 2026 — Jun 2026Seoul, S. Korea

SK Telecom

Sovereign AI Foundation Model Intern · Mentor: Seongmin Ok

  • Built a synthetic data pipeline for tool-use post-training, generating high-quality and reliable trajectories.
May 2025 — Oct 2025Seongnam, S. Korea

Naver Cloud

Foundation Model Research Intern · Mentor: Hangyeol Yu

  • Developed a scalable pre-training synthetic data pipeline to generate large-scale, knowledge-rich datasets for foundation model training.
  • Investigated efficient LLM architectures with strong length generalization and stable performance beyond training context lengths.
  • Explored distributed training frameworks and optimization techniques for large-scale foundation model training.
02

Projects

Omnimodal Foundation Model · SKT

A.X K2 — Sovereign AI Foundation Model

National Flagship AI Development ProjectMar 2026 – Jun 2026

  • Designed and implemented an execution-driven synthetic data pipeline for enhancing tool-use reasoning and agent capabilities.
  • Established strict heuristic filtering and rubric-based validation frameworks to improve data reliability and mitigate hallucinations.
  • Built MCP integrations using K-Skill and public tools, including Korean-Law-MCP, to expand practical tool-use capabilities.
HyperCLOVA · Naver Cloud

HyperCLOVA X — Sovereign AI Foundation Model

National Flagship AI Development ProjectMay 2025 – Oct 2025

  • Developed large-scale, high-quality datasets spanning diverse domains to improve knowledge coverage and generalization.
  • Curated domain-specific datasets in specialized areas, including finance and nuclear power, to strengthen expert-domain capabilities.
  • Designed synthetic reasoning datasets to improve mathematical reasoning and problem-solving performance.
ADD · National R&D

AI Agents for CBRN Operational Command & Guidance

PRML Lab, Korea UniversityDec 2024 – May 2025

  • Contributed to project proposal writing and early-stage planning.
  • Evaluated and deployed serving frameworks for on-premise environments.
  • Developed demonstration systems to validate and showcase the initial service vision.
Industry–Academia

AI Assistant for Elementary Teachers using LLMs

i-Scream Edu · PRML Lab, Korea UniversitySep 2023 – Jan 2024

  • Built high-quality mathematics and science datasets.
  • Constructed short-answer and open-ended benchmarks for elementary education.
Naver Connect · AI Tech

AI Chatbot for Food Delivery Recommendations & Review Generation

Boost Camp AI Tech — Final ProjectNov 2021 – Dec 2021

  • Information retrieval from a large-scale database using Elasticsearch.
  • Devised a purification strategy for harmful data.
  • Constructed a restaurant metadata dataset via web crawling.
03

Publications

International Conference

arXiv · Preprint

04

Awards

Conquer Health: Foundation Model Hackathon for Medical Science

Hosted by Lunit

Developed a public healthcare chatbot that combines a medical science foundation model with medical MCP tools to answer health questions accurately and explain them clearly. Built a framework that surpassed frontier models in accuracy while improving expertise, usability, and real-world effectiveness.

Frontier Award
05

Certifications

Engineer Information Processing

Human Resources Development Service of Korea

Sep. 11, 2026

Domestic
06

Education

Korea University

M.S. in Artificial Intelligence · Advisor: Prof. Seong-Whan Lee

Aug 2023 – Feb 2026 · SeoulGPA 4.31 / 4.5

Researching Context Compression for Large Language Models.

Sogang University

B.S. in Mathematics & B.A. in Economics

Mar 2015 – Aug 2023 · SeoulGPA 3.53 / 4.5

Dual degree in quantitative and analytical foundations.

07

Skills

Programming

PythonStrong

AI model development (PyTorch) and backend engineering (FastAPI).

C / C++Intermediate

Developed through computer science coursework and systems programming.

Languages

KoreanNative
EnglishIntermediate
Focus areasResearch

Foundation models · synthetic data pipeline · agents.

08

Activities

Oct. 2026 — S. Korea

Hugging Face Krew

Member

  • Created the Korean translation of smol-training playbook (huggingface-KREW/smol-training-playbook-ko)
Sep. 2026 — Nov. 2026S. Korea

tensor2slicon

Member

  • Study and analyze the end-to-end execution pipeline from computation graphs and kernels to compilers, runtimes, and chip architectures.
  • Explore memory movement, tiling, the Roofline Model, and GPU kernel programming with Triton.
  • Developed open learning materials with reproducible code, diagrams, and exercises, while laying the groundwork for a future Legato user community.
Jun. 2025 — Aug. 2025S. Korea

Naver Connect — Boost Course Study Coach

Freelance · Study Coach

  • Mentored Pre-Boost Course participants on AI, machine learning, and LLM questions.
  • Presented a live lecture on Transformer architectures.
  • Reviewed and gave feedback on participants' prompting strategies and prompt-engineering assignments.
Jul. 2023 — Aug. 2023S. Korea

LG Aimers 3rd

Member

  • Competed in both the online preliminary round and on-site final hackathon for product sales forecasting.
  • Studied time-series forecasting methods, including D-Linear and related architectures.
Jul. 2021 — Dec. 2021S. Korea

Naver Connect — Boost Camp AI Tech 2nd

Member

  • Studied fundamental AI concepts and methodologies for AI engineering.
  • Improved model performance through competitions in image classification, KLUE, and MRC.
  • Strengthened core engineering skills through collaborative team-based projects.