eonik
AI & ML interests
On-device video intelligence for performance ad production — SigLIP2 Core ML footage search, Whisper transcription, Apple Vision. Multi-provider generative video assembly (Veo, Seedance) with human-in-the-loop QA. MCP agents for desktop video editing workflows.
Recent Activity
eonik
We build eonik, a macOS video editor for teams that produce performance marketing creative at scale. The app combines timeline editing with on-device machine learning and generative video tools so editors can search footage, assemble variants, and export platform-ready ads from their own media library.
This organization hosts open artifacts from that work: model conversions, evaluation fixtures, and schemas we use in production and that other developers can reuse.
On the Hub
| Models | Core ML builds for on-device multimodal search, starting with SigLIP 2 image/text encoders optimized for Apple Silicon (ANE, runtime download, SHA-pinned manifests). |
| Focus | Private, local inference for video production: visual retrieval, transcription, and agent-driven editing workflows via MCP. |
We publish when an artifact is tied to a shipped or shipping product feature, with model cards that document conversion parity, licensing, and usage constraints.
Product & code
- eonik.ai: editor, docs, and downloads
- GitHub · eonik-ai: MCP server and open tooling
- Chrome extension: save ads and references to your research library
- OpenClaw skill: creative audit and planning via MCP
Contact
Integration questions or model issues: connect@eonik.ai