Gian Karl
Colinares

Role: Data Scientist

Analyzed smartwatch-generated stress data to identify patterns and correlations between stress levels and daily activities. Generated visual insights through time-series analysis to identify effective stress-relief activities and peak stress periods.

Role: Data Engineer

Collected and structured heat index datasets by scraping PAGASA benchmarks, LGU Facebook pages, and X sources to build a clean dataset for model training. Enabled reliable heat index forecasting across 23 PAGASA stations with strong performance metrics (RMSE = 2.35°C, MAE = 1.74°C, R² = 0.82).

Role: Machine Learning Engineer

Designed and trained a custom Convolutional Neural Network (CNN) to classify art styles using the WikiArt dataset. Achieved 63.08% classification accuracy across multiple art styles through systematic model tuning and dataset preprocessing.

Role: Machine Learning Engineer

Developed and implemented reinforcement learning agents including Q-Learning, Monte Carlo, and a custom Dyna-Q model to simulate intelligent gameplay strategies in Integer Damath. Created the Dyna-Q algorithm from scratch and evaluated algorithm efficiency and learning performance to compare learning speed and decision accuracy across the three RL approaches.

Role: Machine Learning Engineer

Built a machine learning model using Random Forest Regression to predict flight arrival delays based on operational and scheduling features. Achieved strong predictive performance through systematic feature selection and hyperparameter tuning.