I'm Nelson. I build tools that make AI systems safer to deploy.

My path here was unusual. I spent seven years in behavioral health research - coordinating peer-support programs across 500+ organizations, publishing on workforce collapse in the APA Psychiatric Rehabilitation Journal, and learning how complex systems break when you push people to perform without infrastructure to support them.

That lesson turned out to be universal. The same failure patterns I studied in human systems show up in AI deployments: drift under pressure, coherence loss at scale, collapse when governance is bolted on instead of built in. So I started building the infrastructure.

What I maintain

navi-sanitize ([repo]) - Deterministic input sanitization for LLM pipelines. Strips homoglyphs, invisible Unicode, template injection, and path traversal vectors. Zero dependencies. Live on [PyPI].

grippy ([repo](https://github.com/Project-Navi/grippy-code-review)) - AI-powered code review with a deterministic security rule engine and codebase knowledge graph. Structured findings, severity scores, pass/fail verdicts. Works with any model.

navi-bootstrap ([repo]) - Spec-driven repo scaffolding that ships CI, security scanning, code review, and release pipelines in a single command. Live on [PyPI].

All under [Project Navi]. I also submitted security contributions upstream - a CVSS 9.1 vulnerability disclosure to OpenHands (PR merged to main) and a contribution to the OpenSSF Scorecard project.

Where sponsorship goes

Sponsoring me directly supports the person writing the code, reviewing the PRs, and maintaining the docs. There's no team behind a logo - it's me, a homelab with an RTX 3090, and a commitment to keeping this work open.

If these tools have saved your team time or you want independent AI security tooling to exist outside corporate gatekeeping, this is how you help it continue.