Muhammad
Nouval Rifqi
Student at Universitas Syiah Kuala
Faculty of Mathematics and Natural Sciences
Department of Informatics
Bio
Muhammad Nouval Rifqi is a student at Universitas Syiah Kuala, Faculty of Mathematics and Natural Sciences, majoring in Informatics. His interests span across machine learning, natural language processing, and computer vision. Passionate about research and real-world application, he continuously works on impactful personal and collaborative projects, particularly in data science and software engineering.
He is actively participating in academic competitions and enjoys sharing knowledge through open-source contributions and writing. His goal is to bridge the gap between advanced research and accessible technology.
Areas of Interest
AI & Machine Learning Modelling
Focused on building and evaluating predictive models using tools such as Python, Scikit-Learn, and PyTorch. Experienced in model evaluation, hyperparameter tuning, cross-validation, and applying techniques like ensemble learning, tree-based models, and deep learning architectures.
Natural Language Processing (NLP)
Specializing in processing and understanding textual data using transformer-based models (e.g., BERT, RoBERTa), traditional NLP pipelines (e.g., TF-IDF, N-gram), and sequence modeling techniques such as RNN and attention mechanisms. Skilled in fine-tuning and evaluating language models for real-world applications.
Computer Vision
Proficient in applying convolutional neural networks (CNN), object detection algorithms (YOLO, Faster R-CNN), and image processing techniques with OpenCV and PyTorch. Also experienced in visual analytics, spatial tracking, and building visual dashboards for performance analysis.
Projects
Trump vs Kamala – Sentiment Analysis
A political sentiment analysis project using YouTube debate comments to compare public perception of Trump and Kamala. Utilizes RoBERTa for context-aware sentiment classification and interactive visualization.
Tech: RoBERTa, Sastrawi, YouTube API, Matplotlib, Python
Undercover Game App
A social deduction mobile game inspired by 'Undercover'. Players take on secret roles as civilians or undercovers, using single-word clues to identify others while hiding their identity.
Tech: Android Studio, Java, XML, Firebase
Voice Chatbot – STT, Gemini LLM, TTS Integration
A voice-based chatbot that converts user speech to text, processes it with Gemini API, and responds back with speech. The interface is built using Gradio for interactive testing.
Tech: Whisper.cpp, Gemini API, Coqui TTS, Gradio, Python
Automatic Text Summarization for Indonesian News
A summarization system for Indonesian news articles using the IndoT5-base model fine-tuned on IndoSum. Evaluated with ROUGE metrics and implemented in Colab.
Tech: IndoT5, IndoSum, Transformers, Google Colab, Python
Food Security Analysis in Aceh Using KNN Classification
A machine learning project analyzing food security in Aceh Province using K-Nearest Neighbors. It classifies regions into six vulnerability levels and provides visual and data-driven policy insights.
Tech: KNN, Scikit-learn, Pandas, Matplotlib, Python
Emotion Classification with Bi-LSTM
A deep learning project to classify emotions such as anger, fear, joy, love, sadness, and surprise from text using a Bidirectional LSTM. The pipeline includes data preprocessing, model training, and evaluation.
Tech: Bi-LSTM, TensorFlow, Keras, Pandas, Python
Experience
08/2023 – Present
Junior High School Robotics Teacher – SMP IT Islam Cendekia Darussalam
Involved in curriculum development aligned with technological advancements, encompassing the design of learning materials in the field of informatics, particularly robotics. Taught robotics subjects to students, focusing on basic programming concepts and the use of Microbits software. Regularly assessed students' understanding and progress in robotics subjects and provided training on how to explain created programs to others.
02/2024 – 07/2024
Database Lab Assistant – Universitas Syiah Kuala
Prepared and conducted database laboratory sessions by drafting materials, organizing practical resources, and ensuring necessary software availability. Guided students through fundamental database concepts and provided technical support during practice. Assisted students in understanding difficult concepts with additional explanations and one-on-one mentoring. Evaluated student competency levels and provided constructive feedback to support skill development.
Achievements & Certificates
January 2025
9th Place - DataVers ANAVA
Organized by Universitas Gadjah Mada - Himpunan Mahasiswa Statistika.
In this competition, we tackled predictive maintenance challenges by leveraging sensor data to classify machine status into Normal, Warning, or Breakdown. With a dataset of 13 million rows and 28 features, we handled class imbalance, outliers, and missing values through preprocessing and applied cost-sensitive learning for optimal F1 performance. Our final model, built with XGBoost and GPU acceleration, achieved an F1-score of 0.4803 and ranked 3rd in the Public Leaderboard and 9th in the Private Leaderboard. Our pipeline integrated model tuning, feature selection, and industrial deployment recommendations to ensure model robustness in real-world conditions.
March 2025
Arkavidia - Datavidia
Organized by Institut Teknologi Bandung - Himpunan Mahasiswa Informatika.
In this national-level competition, we were challenged to forecast the daily prices of 13 essential food commodities across 34 Indonesian provinces. Our solution utilized XGBoost for multivariate time series forecasting with advanced preprocessing steps including missing value imputation, outlier detection using IQR/Z-score, and rolling median smoothing. We achieved a mean MAPE under 5%, with beef showing the lowest error (0.86%). Our model handled volatile commodities like chili and garlic effectively by incorporating lag features and spatial patterns, showcasing real-world applicability in economic forecasting. We placed 73rd out of 230 teams.
April 2025
Gammafest 2025 - Data Science Competition
Organized by Himpunan Mahasiswa Statistika IPB University
This competition challenged participants to develop a paper recommendation system by predicting citation links between scientific papers in the mythical Elbaf Library. Our solution involved binary classification using logistic regression and ensemble learning (VotingClassifier), combining Logistic Regression, XGBoost, and Decision Trees. We engineered TF-IDF representations from titles, abstracts, and keywords, and built similarity features such as cosine distance, length ratios, and keyword intersections. Our model was trained using stratified 5-fold cross-validation and optimized for the MCC (Matthews Correlation Coefficient) metric, reflecting robust handling of imbalanced citation data.
April 2025
FIND IT 2025 – Data Analytics Competition
Organized by Department of Electrical and Information Engineering, Universitas Gadjah Mada
In this competition, we developed a machine learning pipeline to assess COPPA compliance risk for mobile apps. We engineered domain-specific features such as privacy compliance scores, child-oriented app indicators, and local developer validation. Our ensemble model combined XGBoost, LightGBM, and CatBoost, each with tailored regularization, and achieved strong AUC scores on validation. Extensive feature engineering included log-scaling of downloads, rating-per-download, region-based privacy laws, and behavioral flags like new apps without privacy links. The final ensemble weighted predictions using model-specific AUC contributions, improving generalization under class imbalance.