AI ENGINEER IN THE MAKING SEOUL, KR

Turning curiosity
into intelligent
systems.

I'm Temuujin — an AI and data science student building useful things at the intersection of research, machine learning, and human experience.

OPEN TO COLLABORATIONSCROLL TO EXPLORE ↓
Temuujin Gerelt-Och working on a laptop
HELLO, I'M
TEMUUJIN.
✳ BUILDING WHAT'S NEXT
37°36' N   127°00' E
01 / SELECTED WORK

Ideas made real.

2024 — NOW ↓

THE PORTFOLIO / 01—04

Selected projects.

From evidence-led research tools to learning how language models work under the hood.

01 / RESEARCH TOOLS / AILIVE ↗

Research GAP

An evidence-aware way to explore a research question. It brings relevant literature, source-linked findings, and possible gaps into one inspectable first pass.

ResearchEvidenceAI
Explore ↗
02 / LANGUAGE MODELS / LEARNINGLIVE ↗

Jarvisn't

An independent small language-model project. I'm building the pipeline from text and tokenization through architecture, training, evaluation, and eventual inference.

LLMsTrainingIn progress
Explore ↗
03 / MACHINE LEARNING / DATACONCEPT

Deal detection system

A system that parses online store items, predicts fair prices with machine learning, and alerts when a listing looks undervalued.

PythonPredictionAutomation
04 / COMPUTER VISION / INTERACTIONCONCEPT

Hand-tracking canvas

A real-time finger-tracking experiment that turns hand motion into on-screen drawing with low-latency landmark detection.

Computer visionOpenCVInteraction

02 / A LITTLE ABOUT ME

Grounded in
learning.
Driven to build.

✳

I'm a student at Kookmin University in Seoul, exploring how intelligent systems can make research and everyday work clearer and more useful.

My interests span machine learning, deep learning, computer vision, and language models. I like understanding the whole pipeline — from data and model training to the experience people actually use.

BASED INSeoul, South Korea
STUDYING ATKookmin University
FOCUS AREASAI · Data Science · Research
View my résumé ↗

TOOLS & THINKING

What I work with.

A growing toolkit for practical, thoughtful AI work — from the first experiment to a more efficient model.

01

Build

Python, JavaScript, Java, C++, Bash, MySQL

02

Model

PyTorch, scikit-learn, neural networks, transformers

03

Explore

Data processing, research synthesis, computer vision

04 / DEEP LEARNING SYSTEMS

Optimize ↗

Exploring faster, leaner training and inference through modern GPU techniques.

Muon optimizerFlashAttention-3Multi-GPU · DDP/FSDPMixed precisionGPU profiling

03 / CONTINUOUS LEARNING

Certificates.

Coursework that helped me turn theory into hands-on practice. Select a certificate to view it.

NEXT / LET'S CONNECT

Have an idea
worth exploring?

Start a conversation ↗