Zain Dana Harper
Kent, Washingtonzaindharper@gmail.comharperz9.github.io
linkedin.com/in/zaindanaharpergithub.com/HarperZ9harperz9.github.io/published-work.html
AI EvaluationSecurity EvaluationPython Developer Tools
Summary
Self-directed builder moving from arboriculture into AI and security evaluation.
Technical Skills
Python, Rust, C++, Dart/Flutter, SQL, GitHub Actions, pytest, Linux, Windows; AI evaluation, planted-fault and control design, red-team and purple-team test cases, MCP, APIs, deterministic testing, offline verification.
Self-Directed Projects
- Shipped Flywheel 1.9.0 on PyPI, a Python/Flutter AI workstation with model routing and a Windows installer; its behavior-check caught 20 of 20 planted changes with 0 of 50 false alarms.
- Built test infrastructure for evaluating AI agents: 324 terminal-state fixtures and 337 component cases that separate failed outputs from provider failures.
- Built Articulate 0.16.0 writing analysis, Accountable Surface approval interfaces, BuildLang 1.10.0 capability checks and Phantom audit and restore.
- Built Daylight Commons 3.2.2 (now CMMNS), a browser extension for nine platforms; slur-sign precision 3,995 of 4,392.
- Ran a pinned METR count_odds task through the Inspect-to-Flywheel bridge with three deterministic controls.
Freelance Technical Writer | Independent projects | Began 2017
- Wrote implementation guides, API documentation, and release notes for developers and customers.
Employment
Arboriculture and Field Operations | Legendary Tree
- Turned site assessments into estimates, proposals, and workable job scopes; coordinated clients, crews, and vendors so teams understood access constraints, timing, and hazards.
Technical Networking Support | Stream/Convergys | 2014 to 2015
- Resolved Xbox connectivity problems across TCP/IP, DNS, NAT, firewalls, and routers; documented troubleshooting steps for consistent support handoffs.
Open-Source Contributions: 37 Merged in 30 Repositories
- DeepEval #2822: fixed crashes on unhashable tool outputs and nested parameters, with regression tests. Merged.
Certifications and Education
- Google DeepMind: Train A Small Language Model, assessed skill badge, 100/100 | Google Skills | September 2026
- Microsoft Applied Skills, four credentials, including Secure AI solutions in the cloud and Azure DevOps pipeline security (95% assessment) | Microsoft | September 2026
- Qualys Certified Specialist, six exams: VMDR, Policy Audit, EDR, TotalCloud, Kubernetes, Custom Assessment | Qualys | September 2026
Training: Cisco Ethical Hacker, Claude Academy MCP, UNESCO Ethics of AI, 17 Linux Foundation security courses. All 42 credentials and verification links: harperz9.github.io/published-work.html
High School DiplomaWilsonville High School, Oregon2013