Skip to content

Robertjam954/multimodal_rag_application / .claude/skills/airunway-aks-setup

airunway-aks-setup

Verified

Set up AI Runway on AKS — from bare cluster to running model. Covers cluster verification, controller install, GPU assessment, provider setup, and first deployment. WHEN: "setup AI Runway", "onboard AKS cluster", "install AI Runway", "airunway setup", "deploy model to AKS", "GPU inference on AKS", "KAITO setup on AKS", "run LLM on AKS", "vLLM on AKS", "set up model serving on AKS", "AI Runway controller".

references/model-sizing.md

View on GitHub
# Model Sizing & Starter Recommendations

## VRAM Sizing Guide

Estimate how much VRAM a model requires. Values are **weights only** — add 20–30% overhead for KV cache and activations during inference.

| Parameters | float16 / bfloat16 | int8 | int4 / GGUF Q4 |
|-----------|-------------------|------|----------------|
| 1B | ~2 GB | ~1 GB | ~0.5 GB |
| 3B | ~6 GB | ~3 GB | ~1.5 GB |
| 7–8B | ~14–16 GB | ~7–8 GB | ~3.5–4 GB |
| 13B | ~26 GB | ~13 GB | ~6.5 GB |
| 34B | ~68 GB | ~34 GB | ~17 GB |
| 70B | ~140 GB | ~70 GB | ~35 GB |

**Example:** 4× A100 80 GB = 320 GB total. Llama-3.1-70B at bfloat16 ≈ 168 GB with overhead — fits across 4 GPUs with tensor parallelism.

## Starter Model Recommendations

| Cluster Capacity | Model | Provider | Notes |
|-----------------|-------|----------|-------|
| CPU-only | `google/gemma-3-1b-it-qat-q8_0-gguf` | KAITO (llama.cpp) | GGUF Q8; runs on CPU |
| 1× T4 (16 GB) | `microsoft/Phi-3-mini-4k-instruct` | KAITO (vLLM) | ~8 GB float16; fits with headroom |
| 1× A10G/L4 (24 GB) | `meta-llama/Llama-3.1-8B-Instruct` | KAITO (vLLM) | ~16 GB bfloat16; gated — needs HF token |
| 1× A100 40 GB | `microsoft/Phi-3-medium-128k-instruct` | KAITO (vLLM) | ~28 GB float16; non-gated; MIT license |
| 1× A100 80 GB / H100 | `meta-llama/Llama-3.1-8B-Instruct` | KAITO (vLLM) | Oversized; upgrade to 70B if more GPUs available |
| 4× A100 80 GB | `meta-llama/Llama-3.1-70B-Instruct` | KAITO (vLLM, TP) | ~168 GB; tensor parallelism; gated |

### Gated Models

These models are **gated on HuggingFace** and require an access token:
- `meta-llama/Llama-3.1-8B-Instruct`
- `meta-llama/Llama-3.1-70B-Instruct`

**Non-gated alternatives** (no token required):
- `microsoft/Phi-3-mini-4k-instruct` (MIT license)
- `microsoft/Phi-3-medium-128k-instruct` (MIT license)
- `google/gemma-3-1b-it-qat-q8_0-gguf` (Gemma license)
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall