Building tools for practical workflows — data, automation, and currently: image classification.
- Image coherence detection – A ResNet-34 model (
coherence_detection) that classifies images as coherent, incoherent, or semi-incoherent for high-volume review of generated content. Future updates planned with a larger dataset. - NSFW detection – A model (
nsfw_detection) geared specifically toward large-scale analysis of images. Designed for the same high-volume review scenarios as the coherence model. - Open-source utilities – Lightweight configurable automation, data parsing, terminal workflows. Simple, modular, documented.
- Functional tool synthesis
- Emphasis on working code and clear usage examples
- Open-source health tools (privacy-first, data parsing)
- Bridging structured data with creative/media applications
- Developer utilities for file management & terminal workflows
- ML evaluation for generation quality (coherence, safety, and beyond)