Open Source
I believe the future of AI is open, local, and accessible. My open-source work focuses on creating tools and frameworks that empower developers to build intelligent systems without relying on walled gardens.
OmniSLM
OmniSLM is an open-source framework designed to streamline the creation of production-ready AI applications powered by small, locally-hosted language models.
Contribution Philosophy
Privacy First
Code should run locally by default. I prioritize architectures that don't require sending sensitive data to third-party APIs.
Developer Experience
Documentation is a feature, not an afterthought. I strive to provide clear examples, comprehensive type hints, and robust testing.
Roadmap & Planned Features
vLLM Native Integration for OmniSLM
Moving beyond Ollama to support high-throughput, paged-attention inference via vLLM directly within the OmniSLM runtime layer.
Visual Agent Builder
A React Flow based graphical interface for connecting OmniSLM agents, tools, and memory stores without writing code.
Spring AI Enterprise Extensions
Publishing open-source Java libraries that add advanced RAG re-ranking capabilities to the Spring AI ecosystem.
Community Goals
My long-term goal is to build a community around OmniSLM that bridges the gap between academic AI research and production engineering. I actively welcome contributions in the form of:
- New Agent Tool integrations (APIs, calculators, shells)
- Vector Database adapters (Milvus, Pinecone, Chroma)
- Performance optimizations for the memory retrieval layer
- Documentation improvements and tutorial creation