Proud to share something I’ve been building from the ground up, Introducing Skyulf.
A self-hosted MLOps platform that gives data scientists, engineers, and small teams full control over their pipelines, models, and infrastructure.
After months of development, I’ve designed and built it end-to-end:
Backend: FastAPI-based modular pipeline system for ingestion → feature engineering → modeling → evaluation.
Frontend: React Flow interface for visual pipeline composition.
Orchestration: Celery + Redis for async task management, with reproducible artifact tracking via joblib.
Infrastructure: SQLite for dev, PostgreSQL for production, container-first deployment.
Skyulf is still in early alpha, but it already runs full workflows locally or on-prem. I’m continuing to evolve it toward an open-source release focused on transparency, flexibility, and control.
GitHub: https://lnkd.in/dwwydVbC
Website: skyulf.com
Would love feedback, contributions, or even just visibility; every star, share, or suggestion helps move it forward.