Deepak Panigrahy

AI Systems Researcher • LLM Energy Measurement • Systems Optimization

About

I am an AI systems researcher focused on measuring and optimizing the energy efficiency of large language model (LLM) systems.

My work explores cross-layer energy instrumentation for AI systems — from hardware energy counters to orchestration behavior in agentic LLM workflows.

Currently building A-LEMS (AI-LLM Energy Measurement System), a research framework that measures the hidden orchestration energy cost of LLM systems.

Research Project

A-LEMS — AI-LLM Energy Measurement System

A cross-layer measurement framework that captures real hardware energy consumption of LLM workflows including:

CPU energy via RAPL
LLM orchestration overhead
Agentic vs linear execution comparison
Energy per token and per task analysis

Live demo:
https://a-lems-dashboard.onrender.com/

Research Focus

AI Systems Optimization
Energy Efficient AI
LLM Orchestration Overhead
Agentic AI Systems
Cross-layer performance measurement

Open Source

GitHub Profile:
github.com/deepakpanigrahy03

Repositories include research prototypes for:

Energy measurement instrumentation
LLM experiment frameworks
AI system benchmarking tools