CHEA ROTHA

Building end-to-end intelligent systems — from raw data to real-world impact

Data Scientist & ML Engineer with an Engineering degree in Applied Mathematics and Statistics from the Institute of Technology of Cambodia (ITC). I specialize in end-to-end ML system development — from feature engineering to scalable model deployment — with real-world impact across energy systems, macroeconomic forecasting, NLP, and policy analytics.

• 🔋 Currently engineering Battery Energy Storage Systems (BESS) data pipelines at SchniecTech Group
• 📈 Built a hybrid inflation forecasting model (XGBoost + LSTM + SARIMAX) for the Cambodian Ministry of Planning
• 🎓 Former Data Science Instructor — taught ML, EDA, and Power BI to the next generation of analysts
• 🌏 Passionate about applying AI to Southeast Asian development challenges

Mar 2023  ──●  First commit — started exploring data science on GitHub
Apr 2023  ──●  Uploaded early EDA notebooks on tourism & agriculture data
Jun 2023  ──●  Published Laptop Price Prediction project (SVR winner!)
Sep 2023  ──●  Hotel Reservation DB — first SQL schema design on GitHub
Jan 2024  ──●  Joined Sunrise Institute — started teaching & documenting
Mar 2024  ──●  Crop Yield Prediction — XGBoost + Random Forest pipeline
Jul 2024  ──●  Cambodia Tourism Forecasting — ARIMA / SARIMA / Prophet
Oct 2024  ──●  Deep dive into NLP — tokenization & Khmer language corpus
Feb 2025  ──●  Ministry of Planning — Inflation Forecasting Thesis begins
Apr 2025  ──●  Hybrid ML deployed: XGBoost + LSTM + SARIMAX live
Dec 2025  ──●  SchniecTech — EMS/SCADA real-time data engineering begins
Apr 2026  ──◉  Today — actively building, learning, and contributing 🚀

Dec 2025 – Present

• Analyzed high-frequency EMS/SCADA time-series data (Active Power, Frequency, SOC, Voltage, Reactive Power)
• Built real-time interactive dashboards for grid and plant performance monitoring
• Conducted anomaly detection for power fluctuations, voltage deviations, and system faults
• Optimized battery charge/discharge cycles by monitoring State of Charge (SOC) patterns
• Automated daily operational reports for engineering and management decision-making

Feb 2025 – Oct 2025

• Engineered a hybrid inflation forecasting system (XGBoost + LSTM + SARIMAX) to enhance national economic projections
• Built end-to-end data pipelines for macroeconomic indicators including cleaning, feature engineering & stationarity testing
• Analyzed global commodity factors (oil, gold) and domestic sector drivers affecting Cambodian inflation
• Delivered interactive dashboards and policy reports adopted in economic planning decisions

Jan 2024 – Feb 2025

• Taught EDA, statistics, ML, forecasting, and data visualization using Python and Power BI
• Designed hands-on mini-projects bridging theory with real-world datasets
• Mentored students on data storytelling and insight communication to technical and non-technical audiences

A hybrid ML system to forecast national inflation with improved accuracy over traditional models.

• Approach: XGBoost + LSTM + SARIMAX ensemble — combining classical time series with deep learning
• Data: Macroeconomic indicators, global oil & gold prices, domestic sector indices
• Impact: Improved forecast accuracy for national economic planning at the Ministry of Planning
• Tools: Python TensorFlow Statsmodels XGBoost Pandas Plotly

Real-time analytics platform for Battery Energy Storage System operations.

• Approach: High-frequency signal analysis (SOC, Frequency, Active Power) with anomaly detection
• Impact: Enabled proactive grid instability detection and optimized charge/discharge efficiency
• Tools: Python Pandas Power BI EMS/SCADA data Matplotlib

Time-series forecasting of tourist arrivals with post-COVID-19 recovery trend analysis.

• Approach: Evaluated ARIMA, SARIMA, Prophet, and LSTM; selected best performer via MSE and residual diagnostics
• Impact: Identified seasonal recovery patterns → delivered data-driven recommendations to policymakers
• Tools: Python ARIMA SARIMA Prophet LSTM Pandas Matplotlib

ML system predicting agricultural yield and recommending optimal crop selection.

• Approach: Random Forest & XGBoost on soil and weather features; recommendation engine built on top
• Metrics: Evaluated using R², MAE, RMSE
• Tools: Python Scikit-learn XGBoost Pandas NumPy Seaborn

Regression pipeline to predict laptop prices from hardware specifications.

• Approach: Compared Linear Regression, Random Forest, and SVR — SVR delivered best accuracy
• Pipeline: Web scraping → EDA → Feature Engineering → Model Training → Evaluation
• Tools: Python BeautifulSoup Scikit-learn Pandas Matplotlib Seaborn

Relational database system for end-to-end hotel operations management.

• Scope: Room booking, client management, staff scheduling, payment confirmation
• Design: ER diagrams, normalized relational schemas, primary/foreign key constraints
• Tools: MySQL SQL ERD Design Relational Modeling

🏫 Institute of Technology of Cambodia (ITC) — 2020 – 2025

Engineering Degree in Data Science · Major: Applied Mathematics and Statistics 📄 Thesis: Analysis and Forecasting of Inflation in Cambodia

🌐 Aii Language Center — 2019 – 2022

English · French (Basic)

 

---

_{"Turning complex data into decisions that matter."}