AI Deployment & Solutions PM
Open to deployment-shaped AI roles
Two years supporting enterprise fintech clients, then six years at MX building the monitoring platform and the AI agents that came after it. Rolled out to 500+ employees and handed to the teams that own them. Now also guiding 190+ product managers through launching AI prototypes of their own.
01 / Method
Six steps, the same every time, whether the customer is a support team, an executive with an unmet need, or a cohort of product managers. The method is what makes the next deployment faster than the last one.
01
Sit with the people doing the work and read what they escalate. Two years on the support desk taught me where the cost actually lands.
02
Discovery with the stakeholder who owns the outcome, up to VP and C-suite. Then a PRD and a spec, sized to the smallest version that moves the number.
03
Rovo agents, JSM automation, prompt systems, working prototypes. I build it myself, so the pilot never waits on an engineering queue.
04
A readiness gate, a controlled beta, a feedback scorecard, and exit criteria written before launch. The builder is never the approver.
05
Rollout, training, and documentation the team and the tool both read, so the standard and the agent can’t drift apart.
06
The owner runs it, changes it, and reports on it without me. Then I take the next problem.
AI SHIFT Method
A discovery-to-deployment loop for shipping AI safely: Specify the problem, Harvest the data, Inspect the current process, Feed and Test the system, and set Telemetry before anything goes live.
5-Gear AI Adoption System
A staged model that standardizes AI discovery, evaluation, and operational rollout across teams, so adoption is designed rather than hoped for.
Builder and gatekeeper
I separate who builds an AI behavior from who approves it for production. Having done both, I design gates that builders don’t route around.
02 / Selected work
What each one was for, how it was built, what changed, and what is still being validated. The status chip is the honest state: shipped, piloted, or designed.
A working prototype built solo in two weeks and run like a product, from discovery to deploy. Built to test a 50% cut in client goal live times and to move Solution Architects from a sales posture into a partnership.
A shareable vision for an agentic AI platform at MX, researched from how the industry is approaching it and iterated until the buildable shape was clear.
The gate that moves a customer-facing AI behavior from build to production: alpha readiness, a controlled Support beta, a feedback scorecard, and a multi-condition exit. Piloted on the MFA Loop Triager.
A capstone gallery built on Cloudflare Workers so incoming AI PM Bootcamp students can see what prior cohorts shipped. One of three core capstone resources, alongside deployment guides and custom AI skills.
03 / AI assets
The AI systems I have built or run. Most are production systems at MX, built on Atlassian Rovo. One is a personal project that spread on its own. Screenshots land here as they are cleared to share.
A zero-setup personal AI knowledge system that onboards itself, scaffolds a custom capture-and-retrieval pipeline tuned per user, and uses cost-aware model routing.
A meta-agent that designs, structures, and audits production-grade Atlassian Rovo agents for non-technical leaders, so they can deploy a working agent in minutes with the platform's governance rules built in.
An autonomous compliance agent, built with Rovo Architect, that investigates a third-party AI vendor across internal tickets, policies, DPAs, and web sources, then produces an evidence-backed risk score and a publish-ready Confluence assessment in minutes.
A Rovo agent that turns a rough idea into a template-compliant Jira Epic, Story, or Task, with sizing rules and RICE scoring built into the flow. Cut ticket-creation time 80% (30+ minutes to under 5) with 100% template compliance by design. Rolled out across TechOps.
A Rovo agent that classifies every new ticket the moment it enters the Product Support queue in Jira Service Management (JSM), using a deterministic three-stage decision framework and returning strict JSON that JSM Automation maps into routing and telemetry fields. Every ticket gets labeled at creation.
04 / Story
Salt Lake City metro · remote, or relocating for the right role
I started on the support desk at MX in 2017, working tickets from fintech and enterprise clients and turning the patterns into requirements for Product. That is still the seat I design from: what breaks for real users, and what it costs them.
In 2019 I spotted a monitoring gap and built the role to close it. I designed MX’s company-wide monitoring and alerting from scratch, wrote the SLIs and SLOs behind it, ran postmortems with engineers, and wrote client-facing RCAs. The status page I drove from concept to company-wide adoption cut related support calls 90%. When the function grew into a dedicated SRE team, I handed it off. Along the way I built the OpenSearch training that let Support and Customer Success troubleshoot without escalating.
Since 2024 the same loop has run on AI. I gathered requirements for TechOps Project Co-Pilot, built it on Atlassian Rovo, and rolled it out across TechOps. I wrote the Pylon AI Lifecycle Process so the person who builds a customer-facing AI behavior is not the person who approves it, then piloted it on a real triager. I run discovery sessions with VPs and C-suite leaders to scope AI against unmet operational needs, and I wrote the AI SHIFT Method and the 5-Gear Adoption System so the rollout is a method rather than a one-off.
Since April 2026 I have also led the capstone for the AI Product Academy’s PM Bootcamp, guiding 190+ product managers through pressure-testing, building, and launching AI-native prototypes.
Off duty I write, record, and produce music as Evening with Crows. I cold plunge, skate, and go to shows.
2017
Software Support Specialist, MX
Front line for fintech and enterprise clients. Turned support patterns into requirements for Product.
2019
Platform Monitoring Technician, MX
Built company-wide monitoring and alerting from scratch. Status page adoption cut support calls 90%.
2020
B.S. Marketing, Utah Valley University
Minor in Business Management.
2024 – present
Product Support Engineer | AI Operations, MX
Built and rolled out Rovo agents, wrote the lifecycle gate, ran executive discovery.
2026 – present
Teacher Assistant & Capstone Lead, AI Product Academy
Contract. Guiding 190+ PMs through building and launching AI-native prototypes.
05 / Vision & values
AI earns trust in production, not in demos. The deployments that hold up are the ones where someone sat with the people doing the work, scoped the smallest thing that would help, ran it against a real gate, and handed it to an owner who can run it without them. That is the job I want: a fresh problem each time, launched and handed off, then the next one.
Evidence over adjectives
If a claim can’t survive an interview question, it doesn’t ship.
Production is the milestone
A demo is a hypothesis. Adoption is the result.
Governance is a feature
The gate that approves an AI behavior is what makes it worth trusting.
Handoff is the finish line
A deployment is done when the owner can run it, and change it, without me.
Cost is a design constraint
The right model for the task, not the biggest one available.
Useful beats impressive
The best AI work disappears into someone’s easier day.
06 / Opportunities
Deployment-shaped AI roles at AI-native or AI-forward companies, where every engagement is a fresh zero-to-one with the handoff built in. Also open to zero-to-one AI Product Manager roles on internal tools and platforms.
Remote first. Relocation, SF and NYC included, is on the table. The fastest next step is an email.
07 / Contact
Roles, projects, pilots. If it involves getting AI out of the demo and into the hands of the people who own the work, I want to hear about it.
Open to deployment-shaped AI roles