I build end-to-end machine learning systems with a focus on production: experiment tracking, CI/CD, containerized deployment, live monitoring, and local-first AI. Currently open to Machine Learning Engineer roles. Based in Palembang, Indonesia.
- Focus areas: MLOps, deep learning, time series forecasting, applied LLMs
- Education: Computer Science - Data Science, BINUS University
- Experience: Ex-intern at Telkom Indonesia - Data Validation and ID/X Partners x Rakamin Academy - Project based Internship
| Project | Summary | Stack |
|---|---|---|
| Local AI Data Analyst | Agentic AI analyst that turns plain English questions into SQL, runs them against a local database, self-corrects failed queries, and returns a table plus an interactive chart. Runs fully local through Ollama, so no business data ever leaves the machine. | Ollama (qwen2.5-coder), SQLite, Streamlit, Plotly |
| MLOps Obesity Classification | End-to-end MLOps pipeline for obesity classification with experiment tracking, CI/CD, containerized serving, and a live monitoring dashboard with alerting rules. | MLflow, DagsHub, Docker, GitHub Actions, Prometheus, Grafana |
| Intel Image Classification | Scene classifier across 6 classes (around 17K images) using MobileNetV2 transfer learning with two phase fine tuning. 93.4% test accuracy, exported to SavedModel, TFLite, and TensorFlow.js. | TensorFlow, Keras |
| Bitcoin Price Forecasting | Multi-horizon Bitcoin price forecaster using a Seq2Seq LSTM with a custom Multi-Head Attention layer and a training loop built from scratch on the TensorFlow low-level API. | TensorFlow, Keras, Pandas |
| Credit Risk Prediction | Credit default risk model on the LendingClub loan dataset, covering EDA, feature engineering, and model training with Logistic Regression and Random Forest. | Scikit-learn, Pandas, NumPy |
Languages
Machine learning and deep learning
MLOps and deployment
Data and cloud
BNSP - Associate Data Scientist, certified by Komdigi
Verified AWS credentials:
