I consider myself a domain agnostic problem solving generalist, currently operating within the data space building a governed streamhouse architecture at Lunar Bank.
- Agentic SDLC: turning agents, prompts, and project opinions into repeatable engineering systems.
- Reproducible environments: using Nix, dotfiles, and small CLIs to make environments portable.
- Streaming architectures: Using technologies such as Fluss, Flink and Starrocks to bridge batch and streaming.
- Practical knowledge capture: keeping notes, operating models, and reusable learning material close to the code.
| Category | Repo | What it is |
|---|---|---|
| Agentic Engineering | nix-agents | Agents packaged as Nix derivations. |
| Agentic Engineering | pi-extensions | Custom extensions for the pi coding-agent environment. |
| Agentic Engineering | open-knowledge | Public knowledge base and learning surface. |
| Developer Tools and Environments | ezgit | A Go CLI for making git repository management easier. |
| Developer Tools and Environments | git-land | Tools for landing changes deterministically against project opinions. |
| Developer Tools and Environments | nix-config | Nix-backed configuration for machines and tooling. |
| Developer Tools and Environments | bootstrap | Bootstrapping scripts and setup glue. |
| Streamhouse | streaming-playground | Experiments with streaming data systems. |
| Streamhouse | fluss | Fork of Apache Fluss, a streaming storage project for real-time analytics. |
| Miscellaneous | kirksw.github.com | Personal GitHub Pages site. |
- Make implicit project opinions explicit, then automate against them.
- Prefer small, composable tools over large systems that hide the important edges.
- Keep environments reproducible so work can move between machines without ceremony.
- Treat AI agents as engineering teammates that need clear contracts, validation, and feedback loops.
- Optimize for useful, shipped learning: notes, scripts, prototypes, and docs should compound into better future work.
- GitHub: @kirksw
- Email: kirk.sweeney@outlook.com




