AI Architect · Governed Agentic AI · Enterprise Systems Architecture

Austin, Texas · Website · Email · LinkedIn · X

I build AI systems where model capability, human authority, infrastructure, and business consequence meet.

My work spans governed agents, AI-ready data platforms, cloud architecture, and decision systems. I make the boundaries explicit: what the AI may propose, who can authorize action, what evidence must hold, and how the system recovers when something fails.

Clarity over cleverness. Correctness over speed. Evidence over hype.

Current: Founder, Apex AI|ML
Prior: Minotaur Consulting · Investment Analysis / Office of the CIO · Wall Street

Six systems, each built around a business constraint and an inspectable control boundary. The repositories contain the implementation, validation, and stated limits; the website connects them through concise architecture briefs.

Separates agent capability, cloud identity, deployment approval, and runtime permissions. Bounded IAM, short-lived GitHub OIDC identity, and independent S3 verification keep deployment from silently becoming authority. The documented live AWS DEV workflow passed 399 tests.

A PostgreSQL-backed paper-execution boundary that checks current policy and fresh evidence before acting. Release validation covers concurrency, retries, worker failure, replay, and compensation—preserving the original decision history through recovery.

The research seat can propose a trade. It cannot send one. A standalone paper-trading console makes Research, Risk, Execution, and Surveillance distinct. Session-scoped controls reject changes to approved terms, duplicate submissions, and operations after a halt.

A reproducible governance workpaper with indicative framework mappings, human acceptance, and separately verifiable evidence archives. Accepted declarations and unresolved control gaps remain visible together; the workpaper supports review without claiming certification.

A synthetic supply-chain prototype for comparing disruptions and mitigation options through explainable scoring. An approval gate blocks unapproved disruptive actions; execution remains explicitly simulated.

Protects financial claims from changes in grain, time, meaning, or authority. Three parallel OpenAI Agents SDK specialists propose claims from scoped synthetic warehouse facts; deterministic controls assess and reconcile them. 35 tests passed, one clean live SDK run completed, and its saved decisions were replayed locally.

Languages: Python · SQL · Bash
AI & Controls: OpenAI Agents SDK · RAG · Structured outputs · Tool calling · Evaluation · Bounded authority · Human authorization
Cloud & Infrastructure: AWS · GCP · Terraform · Docker · Kubernetes · IAM / OIDC · CI/CD
Data & Systems: PostgreSQL · DuckDB · Linux · Ubuntu · Rocky Linux · RHEL
Domains: Financial Services · Energy & Commodities · Supply Chain · Data Platforms · AI Infrastructure

Engineering principle: Standard library first. Add a dependency when the cost of building exceeds the cost of owning it.

I look for the decision underneath the technology:

• What problem are we actually solving?
• What authority should the AI have?
• What evidence must be true before action?
• What happens when the model is wrong?
• How will we observe, verify, and recover?
• What business consequence does the architecture create?

AI transformation changes workflows, decisions, accountability, and operating models. Those choices belong in the architecture from the beginning.

I write about AI, cybersecurity, risk, and technology at Cyber Essentials and build analytical models that connect technical systems to business decisions.

The Cognitive Convergence — Where Capital Learns to Think explores how AI capability, infrastructure, energy, investment, and orbital systems may reshape capital allocation. Its embedded information layer exposes classification rules, metrics, assumptions, and decision logic.

The conclusion should never stand alone. The evidence and methodology required to evaluate it should travel with it.

Additional visual intelligence:

• RAG Attack Surface — how vulnerabilities propagate across retrieval-augmented AI systems
• Infrastructure / FinOps — architecture, cost, risk, and investment consequence
• Energy / Commodities — decision intelligence across operational and market systems

Explore the Tableau portfolio

Senior opportunities across AI Architecture · Enterprise AI · AI Platforms · Cloud / AI Architecture · AI Transformation Consulting · Technical Chief of Staff.

Based in Austin and willing to relocate within Texas, including the Texas Triangle. Open to in-office or hybrid work and meaningful business travel.

Senior. Technical. Strategic. Evidence-driven.

Close enough to the work to build and reason. Senior enough to shape the direction.

^{Updated October 2026}