using-ml-production

# Using ML Production

## Overview

This meta-skill routes you to the right production deployment skill based on your concern. Load this when you need to move ML models to production but aren't sure which specific aspect to address.

**Core Principle**: Production concerns fall into four categories. Identify the concern first, then route to the appropriate skill. Tools and infrastructure choices are implementation details, not routing criteria.

## When to Use

Load this skill when:
- Deploying ML models to production
- Optimizing model inference (speed, size, cost)
- Setting up MLOps workflows (tracking, automation, CI/CD)
- Monitoring or debugging production models
- User mentions: "production", "deploy", "serve model", "MLOps", "monitoring", "optimize inference"

**Don't use for**: Training optimization (use `training-optimization`), model architecture selection (use `neural-architectures`), PyTorch infrastructure (use `pytorch-engineering`)

---

## How to Access Reference Sheets

**IMPORTANT**: All reference sheets are located in the SAME DIRECTORY as this SKILL.md file.

When this skill is loaded from:
  `skills/using-ml-production/SKILL.md`

Reference sheets like `quantization-for-inference.md` are at:
  `skills/using-ml-production/quantization-for-inference.md`

NOT at:
  `skills/quantization-for-inference.md` ← WRONG PATH

When you see a link like `[quantization-for-inference.md](quantization-for-inference.md)`, read the file from the same directory as this SKILL.md.

---

## Routing by Concern

### Category 1: Model Optimization

**Symptoms**: "Model too slow", "inference latency high", "model too large", "need to optimize for edge", "reduce model size", "speed up inference"

**When to route here**:
- Model itself is the bottleneck (not infrastructure)
- Need to reduce model size or increase inference speed
- Deploying to resource-constrained hardware (edge, mobile)
- Cost optimization through model efficiency

**Routes to**:
- [quantization-for-inference.