Tempe, Arizona, United States
I'm a senior software engineer with 8 years across distributed back-end systems, internal platforms and developer tooling, and full-stack product — currently an AI/ML Senior Development Engineer at Curana Health and a PhD candidate at Arizona State University researching human–agent interaction: data visualization, HCI, agentic-workflow interaction, and engineering ethics. My focus is the seam between the two: building agentic AI systems that aren't just fast and reliable, but **legible and steerable** for the people who depend on them. That's also my research — making agent behavior observable and supervisable as AI takes on more of the work. Right now I'm building **Foreman**, an open-source environment for supervising multi-agent software development — moving human oversight of AI coding agents from *reading what they did* (transcripts + diffs) to *inhabiting where they are*: a real-time view where a supervisor both observes and intervenes. What I bring: years of shipping real software, plus research depth in exactly the problem the field is racing to solve — keeping humans in command of increasingly autonomous agents. Interests: agentic systems · AI/agent observability & evals · LLM tool-use · distributed systems.
• Built an internal platform that makes autonomous coding-agent sessions observable, durable, and team-owned — re-homing each session from a single laptop to the workload it's acting on (a ticket or branch), as an append-only log of turns with a real-time dashboard and an "attention" model that surfaces exactly where an agent needs a human. • Built a production multi-agent RAG assistant on Microsoft Teams — an orchestrator + sub-agent architecture with parallel tool dispatch and full OpenTelemetry observability, plus an automated eval harness that grades the live bot with LLM-as-judge scoring and auto-files improvement tickets. The multi-agent rewrite ran faster than the single-agent baseline at equal quality. • Built a custom prompt/model/parameter optimization engine — an evolutionary search over (prompt, model, parameters) against a labeled corpus, with a three-primitive per-field evaluator (exact-match, embedding-cosine, LLM-judge) and a frozen-holdout guard against metric-gaming. • Turned Claude Code into shared team infrastructure with a context-engineering layer — domain knowledge encoded into specialized, context-scoped subagents plus headless multi-repo agent orchestration, so the team's agents behave reproducibly. • Sole architect of the team's HIPAA/SOC 2 Azure platform — a from-scratch, four-subscription landing zone with zero-secret OIDC identity, an AI gateway that centralizes model routing and logs per-call bias and accuracy signals, and a one-command workload-onboarding factory.
• Researching human–agent interaction at the Sonoran Visualization Lab (advisor: Dr. Chris Bryan): data visualization, HCI, agentic-workflow interaction, and engineering ethics — focused on making agentic AI workflows legible and supervisable as AI takes on more of the work. • Building an open-source agent-supervision and observability environment that moves human oversight of AI coding agents from reading what they did (transcripts and diffs) to inhabiting where they are — a real-time view where a supervisor both observes and intervenes.
• Deployed cloud-based remote development environments on AWS — remote coding hosts that gave engineers consistent, ready-to-use dev machines. • Built a CI/CD and DevOps pipeline for a large defense prime contractor, automating build, test, and deployment.
• Developed automated test software for RF (radio-frequency) hardware modules — test harnesses and instrumentation for validating module performance. • Built data-visualization platforms that turned test and telemetry data into interactive views for engineering analysis and debugging.