Applied ML · Data Systems · AI Automation
MSc Electrical & Computer Engineering @ University of Alberta
I build data-intensive systems across machine learning, automation, and scientific computing. My work emphasizes reproducible pipelines, controlled evaluation, and cross-level system validation.
MethaneFuse — Applied ML
Partial-observation multimodal methane detection using Sentinel-2, Landsat 8/9, EMIT, and Sentinel-5P. PyTorch training and evaluation pipeline; reported 480 m classification result: AUROC 93.62, F1 84.87.
MethaneUnion — Data Engineering
Multi-source satellite dataset construction with spatial/temporal matching, quality filtering, and event-level temporal/geographic splits designed to prevent leakage. Contains 8,981 observable multi-sensor events.
MSRE-RT — Systems & Validation
Python numerical reference, C++ implementation, and FPGA/Vitis HLS integration with cross-level verification. Measured 3.043 ms execution for a 1 s simulation step—329× faster than real time.
Jobops — AI Automation
Engineering prototype for deterministic workflow orchestration, durable state, controlled AI use, and browser automation. Human-in-the-loop safeguards enforce explicit submission and verification boundaries.
Data & ML: Python, SQL, Pandas, PyTorch
Systems: C++, Linux, FPGA / Vitis HLS
Application systems: FastAPI, PostgreSQL, browser automation
Engineering: Git, testing, reproducible experiments