Infrastructure capacity forecaster
Capacity planning at your company is one senior engineer's spreadsheet and a lot of intuition. Design an agent that forecasts infrastructure needs from usage curves and upcoming launches. Forecasts are always wrong at the tails — so how does the agent express uncertainty honestly, which decisions may it feed directly versus which need a human review of the forecast, and how would you evaluate forecast quality over time? Finish with how you'd pitch this to the infrastructure director.
score_band_calibration(low: list[float], high: list[float], actual: list[float], target_pct: int) → list[str][[80,90,100,110,120,130,140,150,160,170],[120,130,140,150,160,170,180,190,200,210],[100,95,135,148,155,131,175,189,199,240],90]out["periods=10","covered=9","coverage_pct=90","under_provisioned=1","verdict=calibrated"][[95,95,95,95,95,95,95,95,95,95],[105,105,105,105,105,105,105,105,105,105],[100,130,99,140,101,160,96,180,104,200],90]out["periods=10","covered=5","coverage_pct=50","under_provisioned=5","verdict=overconfident"][[],[],[],90]out["periods=0","covered=0","coverage_pct=0","under_provisioned=0","verdict=insufficient_data"]Turn the fuzzy goal into a bounded agent task. Say what the agent reads, what it may do on its own versus draft for a human, and what it must never touch. Then earn the trust: how you’d verify it before rollout, and the pitch (the problem, what you built, why it matters).
Agent build: design the agent in plain language (or narrate it). The coach grades the decomposition, the guardrails, the verification plan, and the pitch.