I'm Utkarsh — an AI/ML engineer at Amazon who trains models, frames photographs, and runs half marathons. Same method everywhere: show up, do the reps, get stronger. This is my training log.
Block 02 — Strength · AI Engineering — 6 sets
Heavy compound work.
The big lifts. Each set is a project — open one and the reps are the deep-dive: problem, idea, build, result.
REP 1 The problem: LLM deployments waste tokens, cost, and carbon.
REP 2 The idea: the first standardized metric for sustainable LLM deployment.
REP 3 The build: an optimizer balancing response quality against token efficiency.
REP 4 The result: lower compute cost and carbon footprint without sacrificing performance.