At DIPPER Lab, I engineer lightweight PyTorch and C++ inference models for low-memory edge devices and build Python pipelines for time-series data. I also created an integration layer connecting a C++ inference engine to a Flutter frontend, enabling real-time offline use for 100+ local users.
I built SmartNitro, a computer vision project that analyzes multispectral crop and soil imagery to generate fertilizer recommendation heatmaps. At Hoot, I created a Python script for product-image metadata extraction and developed Next.js and TypeScript applications; at easy Go, I refactored MySQL tables and implemented reusable file-upload infrastructure.
