All Work

AI Production Engineer

ComfyUI Production Pipelines

254-node workflows, documented GPU benchmarks, and a 7-color node coding system for headless batch production.

Context

ComfyUI as generation tool vs ComfyUI as production system are different things. Production requires reproducible pipelines, headless execution, documented costs.

Strategy

GPU benchmarking, node-based workflow design, API integration for batch automation.

Solution

Production infrastructure across consumer and cloud GPUs with standardized coding system.

GPU Benchmark

13 GPU types on RunPod. Price-per-clip metric. RTX 5090, L40S, A100, H200. VRAM ceilings documented.

Workflows

254-node face swap (WAN Animate + VitPose + YOLO). Full influencer pipeline (7 stages with checkpoints). Headless batch via ComfyUI API.

Node Coding System

7-color system for workflow legibility. New operator traces execution logic in under 5 minutes.

13
GPUs benchmarked
254
Node pipeline
7
Color coding system
ComfyUI RunPod LoRA GGUF WAN 2.2 Flux ESRGAN Batch Automation