Muhammad Yusuf Haikal

Developed a Next.js-based AI report generator for Klinik Pintar, utilizing RAG with PostgreSQL, pgvector, and Google text embeddings to enable business analysts to create schema-driven reports with high accuracy, featuring secure API key management and multi-provider support (Google Gemini, DeepSeek).

Applied CRISP-DM to analyze NASA VIIRS fire hotspot data, using K-Means, DBSCAN, and Hierarchical Clustering to identify high-risk zones and inform fire mitigation strategies in Indonesia.

Developed a time series model using LightGBM to predict food commodity prices across Indonesian provinces, integrating Google Trends and weather data with Combined Linear-KNN imputation.

Built an ensemble ML model (XGBoost, LightGBM, CatBoost, NN) for FIND IT! 2025 to predict COPPA risks, placing 10th out of 227 teams with SMOTE and preprocessing.

Analyzed 33,140 records from the Global Terrorism Database, built ML models to predict attack targets and damages, and applied clustering with K-Means and Hierarchical Agglomerative methods.

Developed an intelligent lighting system using LDR and PIR sensors with XMEGA-A3BU XPLAINED microcontroller to adjust lighting based on ambient light and human presence.

Led a team to design Tanam, a digital solution for farmers to adapt to climate change, using user research, personas, and high-fidelity prototypes for an intuitive UX.

Built Pacilflix, a streaming platform with PostgreSQL, featuring user authentication, show browsing, reviews, downloads, and admin functions for content and subscription management.