A selection of projects that demonstrate my passion for data-driven solution and engineering.
A study assessing how weather variables alone can predict fine particulate matter (PM2.5) levels in Yogyakarta using ensemble ML algorithms, highlighting key influencers like air pressure and solar radiation.
In this project, I developed a predictive model to estimate sentence lengths in months from raw Indonesian Supreme Court verdict texts, using advanced NLP techniques and ensemble machine learning models to handle noisy legal data.
Developing an advanced NLP pipeline to classify multilingual user reviews of AI chatbots like GPT, Gemini, Claude, and Grok.
Developed a machine learning system to predict citation links between scientific papers using advanced feature engineering, NLP embeddings, and ensemble models, achieving strong performance in a competitive challenge.
A multi-task learning model using EfficientNetB7 to classify clothing type and color, scoring 0.97 Exact Match Ratio on the public leaderboard.
Developed a Convolutional Neural Network using CSRNet for accurate crowd density estimation in images, optimizing for computational efficiency in a competition setting.
Developed an advanced driver drowsiness detection system using state-of-the-art video masked autoencoders, achieving top F1 scores in the Data Slayer 3.0 machine learning competition through innovative preprocessing and model adaptation techniques.
Developed a machine learning model using EfficientNetB7 to classify static images as fall or non-fall events, leveraging computer vision and multitask learning for enhanced safety applications, particularly for vulnerable groups like the elderly.
In this project, I developed a weighted ensemble model to forecast prices of 13 essential food commodities across 34 provinces in Indonesia, leveraging time series data from 2022-2024 and achieving a low MAPE of 3.28%.
