Predictive Sales Forecasting
End-to-end ML pipeline that forecasts retail sales using LightGBM and SARIMA ensembles, reducing MAE by 18% over baseline. Includes automated feature engineering and a FastAPI serving layer.
Hello, I'm
Ariansyah
|
Turning data into insight and equations into solutions.
About Me
I'm a data scientist and applied mathematician with a passion for extracting meaning from complex datasets and developing elegant mathematical models that solve real-world problems.
My work sits at the intersection of statistics, machine learning, and numerical analysis. I enjoy everything from exploratory data analysis to building and deploying production-ready ML pipelines.
Outside of work I love reading papers on optimization theory, contributing to open-source scientific Python libraries, and teaching math concepts through visualizations.
Hire MeSkills & Tools
Data Science
- Exploratory Data Analysis
- Statistical Modeling
- Time Series Analysis
- Data Visualization
Machine Learning
- Supervised & Unsupervised Learning
- Deep Learning (PyTorch)
- Model Interpretability
- Feature Engineering
Applied Mathematics
- Numerical Optimization
- Linear Algebra & Calculus
- Probability & Stochastic Processes
- Differential Equations
Languages & Tools
- Python · R · Julia
- SQL · Spark · dbt
- NumPy · Pandas · Scikit-learn
- Git · Docker · Jupyter
Python
PyTorch
Scikit-learn
NumPy
Pandas
R
Julia
SQL
Spark
Jupyter
Docker
Git
Featured Projects
Numerical ODE Solver Visualizer
Interactive browser-based visualizer for comparing Euler, Runge-Kutta, and Adams-Bashforth solvers on user-defined ODEs. Built with Julia (DifferentialEquations.jl) and a React front-end.
Bayesian A/B Testing Dashboard
A Streamlit dashboard that performs Bayesian A/B tests using conjugate Beta-Binomial models, providing credible intervals and expected loss curves to support decision-making without p-value misuse.
Graph Neural Network for Citation Prediction
Implemented a GraphSAGE model in PyTorch Geometric to predict citation counts of academic papers from their abstract embeddings and co-authorship graph structure, achieving top-5% on the OGB benchmark.
Stochastic Gradient Descent Visualizer
Real-time 3-D loss-landscape visualization of SGD, Adam, and RMSProp on user-configurable benchmark functions (Rosenbrock, Rastrigin, etc.). Pure JavaScript with Three.js for 3-D rendering.
Geospatial Clustering of Urban Mobility
Applied HDBSCAN and DBSCAN to GPS traces from ride-share data to identify mobility hotspots and transit deserts at city scale. Results presented as interactive Folium maps.
Education & Experience
2024 – Present
M.Sc. Applied Mathematics
University of Excellence
Focus on numerical methods, stochastic analysis, and scientific computing. Thesis: Adaptive step-size methods for stiff SDEs.
2022 – 2024
Data Scientist – Analytics
Tech Corp, Remote
Built and maintained ML models in production; reduced churn by 12% through propensity scoring; mentored two junior analysts.
2019 – 2022
B.Sc. Mathematics & Statistics
State University
Graduated cum laude. Minored in Computer Science. Capstone project: dimensionality reduction for high-dimensional genomic data.
Get in Touch
Have a project in mind, a dataset that needs taming, or just want to talk math? Drop me a message!