I'm currently working as an Undergraduate Research Assistant at the Computer Architecture Lab, focusing on optimizing 2D image convolution for NVIDIA RTX 4090 hardware.

Basically, I'm finding ways to make it faster. I started by getting some baseline measurements with standard C++ code so we know what we're improving on. Now, I'm working on specific GPU memory optimizations, like using shared memory tiling and constant caching, to cut down on latency. I'm also digging into the hardware details with Nsight profiling tools to watch things like memory throughput and warp occupancy, which helps me fine-tune the performance.