Gian Bernales

I am a dedicated professional focused on the seamless integration of machine learning into scalable web solutions. My experience spans the full development lifecycle, including front-end design, server-side integration, and database management. This background demonstrates a readiness to tackle diverse tasks and acquire new skills in a high-speed setting.

As a flexible team player, I prioritize rapid learning and critical thinking to solve evolving technical hurdles while bridging the gap between user-facing design and backend infrastructure.

HR Technology Intern - UnionBank of the Philippines - Pasig City, Philippines

(2026-04 - 2026-06)

• Conducted User Acceptance Testing (UAT) for newly deployed Biometrics and QR Scanner validation system updates, and executed test scenarios for the primary financial tracking system's Pension module.
• Engineered multi-criteria lookup formulas to aggregate cumulative asset acquisitions, and deployed optimized calculation formula logic to minimize system latency during heavy dataset operations.
• Audited parallel datasets to isolate record anomalies, effectively identifying system value errors, ledger variance discrepancies, and over-withdrawal calculation defects.
• Collaborated in debugging efforts by logging system enhancement tickets for multi-transaction web UI auto-summation, and documented critical environment blockers following mobile application server rollbacks.

Researcher & Developer - University of Santo Tomas - Manila City, Philippines

(2025-01 - 2025-11)

Temporal Action Recognition System - Undergraduate Thesis

• Co-authored and developed "Shoplifting Recognition through Long-range Temporal Modeling using High-Resolution Network and Encoder-Decoder Temporal Convolutional Networks."
• Focused on Temporal Action Segmentation, utilizing ED-TCNs to capture long-term dependencies in video data for behavioral analysis.
• Demonstrated proficiency in advanced Computer Vision techniques, moving beyond standard object detection to complex behavioral understanding.
• Led the preprocessing phase by converting unstructured video data into normalized keypoint sequences, effectively reducing model noise and improving the system's ability to distinguish between normal shopping behavior and suspicious concealment.

Bachelor of Science - Computer Science - University of Santo Tomas (2023-05 - 2026-06)

Bachelor of Science - Information Systems - University of Santo Tomas (2022-08 - 2023-05)