Projects

• Energy- and Thermal-Aware Scheduler for Heterogeneous Multicore Systems (2022–Present)
  Developed a Hierarchical Multi-Agent Reinforcement Learning (HMARL) framework for OpenMP DAG workloads, achieving 31.7% energy reduction and 34.1% makespan improvement on Jetson TX2 and Intel Core i7. Published in RTSS 2024 (WIP) and submitted to EMSOFT 2025.

• Feature-Aware Task-to-Core Allocation (2023–2025)
  Designed a statistical learning framework using Random Forest, backward stepwise regression, and Pearson correlation, reducing energy by 10% and temperature by 5°C. Published in RTCSA 2025.

• Distribution-Aware Flow Matching for Few-Shot RL (2023–2025)
  Proposed a flow-matching approach for DVFS in few-shot RL, improving frame rates by 30% in neuromorphic vision workloads. Submitted to ECAI 2025.

• FPGA-Based Optical Flow Calculation (2020)
  Implemented a parallel event-based histogram on FPGA for optical flow calculation, published in ASAP 2020.

• Neural Network for Human Activity Recognition (2015)
  Designed a statistical-based neural network for human activity recognition, published in International Journal of Computer Applications.