Srisukanya Gajula

Ai Research Assistant @Briviant Inc

Irvine, CA, US

Benchmarked CNNs (ResNet, AlexNet, SpiralNet) using PyTorch and Lava; documented tradeoffs and performance trends- Used PyTorch Profiler to analyze runtime and apply model optimizations (quantization, schedulers, regularization)- Reverse-engineered SpikeSim latency code to match theoretical hardware-aware models; implemented pipelining and architectural constraints- Adapted and validated custom latency models for spiking transformers based on research papers and GitHub repositories