# quinnjr.dev

> The personal site of Joseph R. Quinn, Esq. — a full-stack software engineer
> and attorney based in Miami, Florida. He builds systems in Rust, Go, and
> TypeScript, works on cloud and Kubernetes infrastructure, and holds a Juris
> Doctor from the University of Miami School of Law. The site hosts his resume,
> open-source work, essays, and a long-form manifesto arguing for small
> language models over frontier-scale AI.

## Pages

- [Home](https://quinnjr.dev/home): Profile, technical skills, and certifications.
- [Resume](https://quinnjr.dev/resume): Professional and educational background.
- [Projects](https://quinnjr.dev/projects): Open-source repositories, generated live from GitHub.
- [Articles](https://quinnjr.dev/articles): Essays and technical writing.
- [The SLM Manifesto](https://quinnjr.dev/slm): Long-form argument against frontier-scale AI.

## The SLM Manifesto

A six-chapter argument that the pursuit of artificial general intelligence
misallocates capital, talent, and electricity, and that small language models
are the better instrument for nearly all real work.

- [Introduction](https://quinnjr.dev/slm/introduction): The case against scale as a strategy: why the pursuit of artificial general intelligence misallocates capital, talent, and electricity, and why small language models are the better instrument for nearly all real work.
- [Definitions](https://quinnjr.dev/slm/definitions): Plain, exact definitions of the terms this argument turns on — large language model, small language model, parameters, inference cost, and fine-tuning — fixed in place before the case continues.
- [Evidence](https://quinnjr.dev/slm/evidence): The empirical test: open-weight labs running on less capital and export-throttled chips, measured against Anthropic, OpenAI, and Google on the same public benchmarks — and what the results say about the trillion-parameter thesis.
- [Costs](https://quinnjr.dev/slm/costs): The physical plant beneath the industry: list prices for servers and racks, the capital cost of a datacenter megawatt, cooling, and storage — the economics of frontier-scale AI counted from the hardware up.
- [Backlash](https://quinnjr.dev/slm/backlash): The human case, made from polling, layoff data, and court records rather than argument: where frontier AI was sold as a substitute for skilled judgment, how it failed publicly, and where judges imposed sanctions.
- [Overkill](https://quinnjr.dev/slm/overkill): Why coding agents like Claude Code and Codex run on general trillion-parameter models rather than coding models, what that costs the developer typing shell commands, and what a purpose-built alternative would look like.

## Articles

- No articles are published yet.

## Elsewhere

- [GitHub](https://github.com/quinnjr)
- [LinkedIn](https://www.linkedin.com/in/quinnjosephr/)
- [Parkinson's Disease Map](https://parkinsons.quinnjr.dev/)

## Notes

- Canonical origin: https://quinnjr.dev
- Sitemap: https://quinnjr.dev/sitemap.xml
- Content is authored by Joseph R. Quinn unless stated otherwise. Attribution
  with a link back to the source URL is requested when quoting.