Home/About

AIsystemsteams can own.

I’m Donald B. Havery, an Applied AI / AI Systems Engineer. I take frontier models and build the system that makes them ship: orchestration, grounding, evaluation, and a runtime you can actually trust.

A systems engineer, not a model researcher.

The distinction matters. I don’t train foundation models, run research, or tune weights. That’s a different discipline. I build around the model: the agentic loops, retrieval, evals, MCP tooling, security checks, and handoffs that turn a raw model into a product a team can run.

The shape of the work is Forward-Deployed: sit close to a real problem, build a thin slice that runs in days, prove it against real data, and harden it into something boring to operate. Models are commodities now. The durable value is in the engineering that sits between the model and a thing you can hand to a user.

I choose the runtime to fit the job. Sometimes that is a hosted frontier model, sometimes local inference, sometimes a hybrid path. The constant is operational discipline: privacy, cost, latency, security, and a system the owner can maintain after handoff.

Four principles I don’t bend on.

They’re the difference between a system that demos and one you can leave running.

01 / DEFAULT

Right runtime for the job

Cloud, local, or hybrid is a design decision, not a slogan. Privacy, latency, cost, and the team’s ability to operate it decide the architecture.

02 / MEASURE

Evals before polish

Non-deterministic systems need tests too. Golden sets and quality gates go in early, so every later change is measured rather than guessed at.

03 / SHIP

A real slice, fast

A thin end-to-end path running in days (real data, real model, real output), so we react to something live instead of arguing over a spec.

04 / HARDEN

Make it boring to run

Graceful fallback, observability and cost discipline. The goal is a system you can leave running unattended, not a demo that needs a babysitter.

The tools I reach for.

Chosen for what ships and holds up, not for what’s trending.

PythonClaudeOpenAIprovider routingMCP ChromaDBembeddings / hybrid searchRAGLLM-as-judge evals pytestCI gatesFastAPISQLiteobservability git hooksscheduled jobsHTML / CSS / JSrelease discipline

Education & certifications.

Education

B.S. Computer Science, Capella University In progress.
ACE-evaluated college credits Earned via Sophia Learning. Transferable college credit, applied toward the degree above.
Evals
Regression gates for model behavior
Security
Controls, audit trails, data boundaries
Handoff
Docs, ownership, maintenance path
15
Anthropic certifications

Have a model-shaped problem? Let’s make it ship.

Tell me the outcome you’re after. I’ll tell you what’s buildable, what it takes, and where the model actually helps.