SOEUK BONDOL — DATA SCIENTIST — CAMBODIA

SOEUK BONDOL.

Data Science × Machine Learning × MLOps

Turning raw data into decisions.
Building models that generalise, pipelines that scale,
and systems that actually ship.

View Projects Contact me

sys.profile() LIVE

roleData Scientist

locationPhnom Penh, KH

focusML · CV · NLP

langPython

projects0

commits∞

mouse_use0ms

skill_index[]

Python

ML/DL

SQL

DevOps

DATA SCIENCE· MACHINE LEARNING· COMPUTER VISION· FASTAPI· PYTHON· SCIKIT-LEARN· MLOPS· DOCKER· NLP· PANDAS· DEEP LEARNING· PYTORCH· DATA SCIENCE· MACHINE LEARNING· COMPUTER VISION· FASTAPI· PYTHON· SCIKIT-LEARN· MLOPS· DOCKER· NLP· PANDAS· DEEP LEARNING· PYTORCH·

Projects

THINGS I'VE SHIPPED.

From Document AI and research papers to real-time distributed streaming —
production-grade architectures, clean code, engineered with scientific intent.

001 ML / Computer Vision

Khmer Number Recognition

A deep learning model for recognising handwritten Khmer numerals. Trained a custom CNN on Khmer digit datasets — tackling one of the hardest OCR challenges in Southeast Asian language processing.

Custom CNN architecture · Data augmentation pipeline · Evaluated on real-world Khmer handwriting samples.

PythonDeep LearningCNNOpenCVOCRKhmer NLP

Try Live Demo GitHub

khmer_recognition.py

import torch
import torch.nn as nn

# Khmer digit CNN
class KhmerCNN(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv = nn.Sequential(
            nn.Conv2d(1, 32, 3),
            nn.ReLU(),
            nn.MaxPool2d(2),
        )
        self.fc = nn.Linear(32*13*13, 10)

    def forward(self, x):
        x = self.conv(x)
        return self.fc(x.flatten(1))

# accuracy: 97.4%

002 DE

Data Engineering / Streaming

Real-Time GDELT Trends Analytics

A high-throughput analytical pipeline processing the global GDELT event stream in real time. Ingests live telemetry streams, aggregates geopolitical sentiment windows, and orchestrates daily workflow tasks.

GDELT Live Streams → Apache Kafka → PySpark Streaming → PostgreSQL → Apache Airflow → FastAPI

Apache KafkaPySparkApache AirflowPostgreSQLFastAPIPython

GitHub Project ↗

003 FIN

Data Engineering / Finance

Stock Streaming Analytics

An end-to-end financial streaming and analytical pipeline. Ingests live market data, processes streaming window metrics, stores structured multi-layer data tables, and serves instant API metrics.

Stock APIs → Kafka Producer → Apache Kafka → PySpark Streaming → PostgreSQL (Bronze/Silver/Gold) → FastAPI → React Dashboard

KafkaPySparkPostgreSQLFastAPIReactData Pipeline

GitHub Project ↗

004 RAG

Research / Document AI

KhmerDoc-Ai

An advanced intelligent document processing (IDP) and layout-aware retrieval system for low-resource Khmer business documents. Implements hierarchical fine-grained layout analysis inspired by the KH-FUNSD dataset (APSIPA 2025) to parse complex structures, extract structured JSON schemas, and enable highly accurate semantic RAG search.

APSIPA 2025 Paper Implementation (KH-FUNSD) · Hierarchical Khmer document parsing · Complex JSON tree extraction · Semantic chunking & layout-aware RAG vector search.

PythonDocument AIRAGLayout AnalysisKH-FUNSDVector DBKhmer NLP

GitHub Project ↗

Stack

THE
TOOLKIT.

Every tool chosen with intent. Zero bloat.

Python

PyTorch

Scikit-learn

Pandas

NumPy

OpenCV

Jupyter

FastAPI

PostgreSQL

SQLAlchemy

Docker

GitHub

Linux

Git

Neovim

Astro

HuggingFace

skill_matrix.json CURRENT

Python Data & ML

Data Analysis Data & ML

Machine Learning ML

FastAPI Backend

Linux / Bash DevOps

SQL Data

Docker DevOps

Deep Learning ML

Contact

LET'S
BUILD
SOMETHING.

Open to data science roles, ML consulting, freelance projects, and interesting collaborations. Based in Cambodia — available remotely worldwide. I reply to everything.

Send me an email soeukbondolcc@gmail.com ↗

soeukbondolcc@gmail.com

↗

Available for opportunities

TypeFull-time · Freelance · Contract

ModeRemote worldwide

TZICT (UTC+7)

Response< 24h