AI/ML●●Solid
Piqc – An open-source GPU waste scanner for LLM inference clusters
One-command GPU waste scanner when Kubecost requires full Prometheus setup.
Solve My ProblemNiche Gem
samhoss93
1111d ago
Back to browse
GitHub Repository
Kubernetes scanner that discovers LLMs running on vLLM and extracts their deployment and runtime facts.
11 starsPython
Piqc – GPU waste scanner for LLM inference clusters
by paralleliq·Jun 2, 2026·3 points·0 comments
AI Analysis
●●SolidSolve My ProblemSlick
Read-only GPU waste scanner finds 20-40% cluster spend waste without agents or sidecars.
Strengths
- •Detects three waste types standard monitoring tools miss: idle allocation, tier misplacement, dark capacity
- •No permanent installation required - runs as a Job, prints results, exits cleanly
- •Dollar-estimate output makes waste immediately actionable for budget discussions
Weaknesses
- •Only surfaces problems without automated remediation or optimization suggestions
- •Competes with established tools like Kubecost and Run:ai with deeper integrations
Category
Target Audience
ML engineers, DevOps teams running GPU clusters
Similar To
Kubecost · Run:ai · VPA
Similar Projects
AI/ML●●●Banger
Browser-Native GPU Sharing
Browser-based GPU cluster for LLM inference with HTTP API and SSE broker coordination.
WizardryZero to OneBold Bet
bilekas
1213d ago
AI/ML●Mid
I reduced LLM inference GPU calls by 94% using semantic routing
94% GPU reduction claim needs verifiable benchmarks to stand out.
Bold BetShip It
kanacki
2115d ago
AI/ML●●Solid
AI/ML benchmark for local LLM inference and XGBoost training on GPU/CPU
One-command benchmark suite comparing Ollama and XGBoost performance with a shared Streamlit dashboard.
Solve My ProblemNiche Gem
albedan
201mo ago
Infrastructure●●Solid
VMetal – run a GPU cloud on bare metal without OpenStack
Saves neoclouds months of engineering by turning bare metal racks into managed Kubernetes clusters.
Solve My ProblemNiche Gem
teb510
1213mo ago
AI/ML●●Solid
WebGPU LLM inference comprehensive benchmark
Sequential-dispatch methodology corrects 20x overestimation in prior WebGPU benchmarks.
Big BrainNiche Gem
yu3zhou4
222mo ago