AI Engineer · Dallas, TX
I build AI systems that hold up in production.
Four of mine run at FieldPulse right now — drafting support email, handling payment notices, answering questions in Slack, and keeping customers updated on their tickets.
Selected work
Five systems in production
What each one does, how it is built, and what it had to get right before it could touch a customer.
Support Inbox Agent
Every support email gets a first draft in seconds. A person still decides what sends.
Internal Knowledge Assistant
1,090 questions answered in Slack, each one cited back to the documentation.
Customer Status Pages
Every customer gets a live page for their tickets. A model writes the updates; five checks decide whether they publish.
Payments Operations Agent
Ten minutes of manual work per payment notice, automated — without the model writing a word of it.
Customer Onboarding Platform
New customers set up their own account instead of waiting on a specialist.
What I bring
Four things that keep a system alive
Getting a model to do something impressive takes an afternoon. These decide whether it is still trusted six months later.
Evals, not vibes
Every system I ship has a way of telling me when it gets worse. Regression suites built from cases that actually happened, and gates I deliberately try to break — a check that has never failed has not been tested.
Fail closed, not open
When money is involved, the question is not whether the model is good — it is what happens when it is wrong. On the payments agent the model never writes the message: it picks a template and fills in the blanks.
People can always stop it
Every draft lands somewhere a person can read it, change it, or kill it before a customer does. On the support agent, auto-send stayed off for the entire first rollout — on purpose. Automation earns trust by being interruptible.
The integration nobody demos
Salesforce, four internal services, three different auth schemes, and an API budget with no room to spare. Plus orchestration that fails softly, so a customer is never stranded halfway through setting up their account.
- 4
- Systems in production
- 1,000+
- Questions answered
- 180+
- Automated tests
- 4
- Services integrated
Got something that has to work?
If you are putting an AI system in front of real users, that is the work I do. Tell me what you are building and where you think it breaks.