Projects
Projects I am proud of
This page highlights three projects that reflect different parts of my interests: machine learning infrastructure, developer tooling, and practical AI systems.
Project overview
| Project | Focus | Technologies |
|---|---|---|
| sbss | Machine learning dataset splitting | Python, NumPy, SciPy, PyPI |
| tau.nvim | Developer workflows and editor ergonomics | Lua, Neovim API, Telescope.nvim |
| pi-chroma | Embeddings and local semantic search | Python, vector databases, AI tooling |
sbss
sbss is a Python library implementing Similarity-Based Stratified Splitting (SBSS) for creating more representative machine learning dataset splits.
- Technologies: Python, NumPy, SciPy, PyPI
- Why it matters: Dataset splitting affects how reliably we evaluate machine learning systems.
The project reflects my interest in machine learning infrastructure, reproducibility, and applied AI research. It explores how better splitting strategies can support more meaningful evaluation workflows.
tau.nvim
tau.nvim is a Neovim plugin/configuration project focused on improving developer workflows and editor ergonomics through Lua-based tooling.
- Technologies: Lua, Neovim API, Telescope.nvim
- Why it matters: A good development environment reduces friction and helps ideas move faster from thought to implementation.
This project explores extensible developer tooling and terminal-native workflows within the Neovim ecosystem. It reflects my interest in productivity systems, customization, and open-source tooling.
pi-chroma
pi-chroma is an experimental project exploring embeddings, retrieval systems, and local semantic search workflows inspired by vector database systems such as Chroma.
- Technologies: Python, vector databases, embeddings, AI tooling
- Why it matters: Retrieval systems are becoming an important part of practical AI workflows.
This project was built as part of my exploration into retrieval-augmented systems and local-first AI tooling. It demonstrates my interest in semantic search, intelligent workflows, and practical AI infrastructure.
What connects these projects?
Although the projects are technically different, they share a common theme:
I enjoy building tools that make research, development, and experimentation more reliable and expressive.
Whether the work involves dataset splitting, editor workflows, or retrieval systems, I am drawn to projects that improve the way people think, build, and evaluate systems.