Projects

Routing AI inference workloads to green modular data centers based on predicted power availability. Explores cross-site routing strategies to minimize carbon footprint of large-scale AI inference.

A prompt-based congestion control framework for Large Language Models. Introduces mechanisms to manage inference request throughput and latency in multi-tenant LLM serving systems.

Resource profiling and energy measurement for next-generation Radio Access Networks. Built open-source tools for real-time monitoring of 5G RAN energy consumption using USRP and containerized deployments.

Implementing convolutional neural networks on FPGA using pipelined FFT architecture for ultra-reliable, low-latency, and energy-efficient inference at the edge.