AI Infrastructure Readiness
Before you can optimize GPU costs or ship reliable inference, you need to know where you stand. This guide covers the diagnostic, the deep dives, and the implementation paths.
The Diagnostic
A fixed-scope audit that turns assumptions into baselines and gives you an executable 90-day roadmap.
AI Infra Readiness Audit
2-3 weeks. Scorecard, cost model, risk register, and prioritized roadmap.
The audit covers GPU cost baselines, reliability gaps, deployment safety, and failure modes. You get a written assessment and a sequenced plan with owners.
Deep Dives
Technical guides covering specific aspects of GPU infrastructure: cost, reliability, architecture, and debugging.
3 min read
AI Infra Readiness Audit: What I Check (and What You Get)
A practical checklist for auditing production AI infrastructure: GPU cost baselines, reliability risks, and an executable roadmap.
4 min read
GPU Cost Baseline: What to Measure, What Lies
Before you can cut GPU costs, you need to measure them correctly. Here is what to track and what the cloud console will not tell you.
5 min read
GPU Failure Modes: What Breaks and How to Debug It
Common GPU infrastructure failures in production and how to diagnose them before they become incidents.
4 min read
Hybrid/On-Prem GPU: The Boring GitOps Path
A practical guide to running GPU workloads on-prem or hybrid, using Kubernetes and GitOps patterns that make operations boring.
4 min read
SLOs for Inference: Latency, Errors, Saturation
How to define meaningful SLOs for production inference workloads, and what to do when they break.
Related Resources
Additional posts on GPU infrastructure, Kubernetes, and inference workloads.
8 min read
Getting Gemma 4 Running on a Radeon 7900 XTX (with and without TurboQuant)
What it took to get Gemma 4 E4B serving cleanly on Radeon through FlexInfer: a stable TRITON lane on a 7900 XTX, an experimental TurboQuant long-context lane on a second node, and the GPTQ pipeline work still underway.
8 min read
Two-Lane Text GPU Allocation: Quality + Vision/Fast (Plus a Media Lane)
How I redistributed 6 models across 3 GPU nodes to eliminate contention, using priority-based shared groups and label-based aliases for routing and failover.
10 min read
Deploying MLC-LLM on Dual RX 7900 XTX GPUs: Debugging VRAM, KV Cache, and K8s GPU Scheduling
What actually broke when I deployed MLC-LLM across two RX 7900 XTX nodes, and the fixes that made it stable: quantization, KV cache sizing, and Kubernetes GPU hygiene.