I’m Petie Clark, a hands-on Principal Applied AI Systems Engineer based in Southeast Georgia.
I build systems that have to do more than produce a convincing demo. My work sits where AI models, enterprise integrations, infrastructure, human judgment, and production operations meet. That includes persistent agent orchestration, evaluation, approval boundaries, release evidence, rollback, and the less glamorous work required to keep a useful system trustworthy after launch.
At Tradeweb Markets, I work in integration engineering inside a regulated financial-technology environment. Through Clearline Tech Methods, the independent applied-AI systems practice I founded, I architect and operate agentic delivery systems, laboratory workflows, managed infrastructure, and production support.
What I build
My most direct applied-AI work is Friday, a governed agentic operating system built around persistent context, specialist routing, bounded authority, independent acceptance, and explicit human approval for consequential actions. It handles hundreds of meaningful investigations, builds, reviews, task executions, and recurring checks in an ordinary week.
I also use governed agent workflows in LIMS delivery. The meaningful improvement is not that a model suddenly knows everything about a laboratory. It is that investigation, implementation, verification, remediation, documentation, and release preparation can move through bounded specialist lanes instead of waiting in one long generalist queue. In practice, that has compressed complex work from months to days while preserving named human accountability.
The surrounding systems work matters too:
- Mutiny demonstrates distributed communications, native and web delivery, voice, file handling, notifications, release engineering, and production operations.
- Blackglass demonstrates security-sensitive native architecture through encrypted messaging, multi-device identity, replay protection, revocation, rollback journals, and fail-closed delivery behavior.
- Enterprise integration work keeps the AI story attached to real identity, workflow, reliability, and operational constraints rather than isolated prototypes.
Neither Mutiny nor Blackglass is an AI product. They are evidence that I know how to build and operate the systems around AI: state, transport, security boundaries, deployment, recovery, and supportability.
How I work
I prefer narrow contracts, one writer per authoritative lane, deterministic checks where judgment is unnecessary, and independent acceptance where the builder should not grade its own interpretation. Agents may investigate, implement, test, and prepare a release. They do not silently widen their authority because the work became inconvenient.
That approach is slower than a demo and much faster than recovering from an unowned production decision.
This blog is where I write about those tradeoffs: what I built, what broke, how authority and failure boundaries were handled, and what the evidence changed.
Outside the terminal
I’m a husband, dad of four, former worship pastor, and committed Christian. Faith and family are not side notes to the work. They shape what I am willing to optimize, which risks are worth taking, and what “better” is supposed to mean.
When I am not building systems, I am usually cooking, reading, working around our small homestead, or explaining why Boba Fett remains the correct answer.
For principal or staff applied-AI systems work, email petie@petieclark.com or find me on LinkedIn.