Show HN: Gemma 4 Multimodal Fine-Tuner for Apple Silicon

Build a unified, memory-efficient data streaming pipeline that allows local fine-tuning of multimodal models like Gemma 4 using large-scale datasets stored in cloud storage. This bridge between cloud storage and local Apple Silicon hardware solves the primary bottleneck for developers with massive datasets and limited VRAM.

Suggested repo: nanoStreamer

"Fine-tune massive multimodal models on your Mac without downloading the entire dataset."

Estimated effort: 40h