I am a software and AI engineer focused on developer infrastructure, application security, AI systems, and automation.
My work turns complex operations into explicit, testable, and auditable systems. I care about what happens outside the happy path: deterministic behavior, honest failure modes, secure boundaries, and evidence that can be independently verified.
|
Trustworthy systems Evidence, policy, reproducibility, and explainable decisions built into the product. |
Developer infrastructure Compilers, control planes, CLIs, APIs, CI/CD, and tools that improve engineering workflows. |
Secure automation Guardrails that preserve human oversight without hiding risk or operational complexity. |
|
Autonomous engineering with verifiable changes. Runs repository checks in isolation, collects security and dependency evidence, evaluates policy, and emits a hashable proof manifest for every change.
|
Accessibility engineering beyond one-off audits. Analyzes real user journeys, turns accessibility findings into reproducible evidence, and helps teams prevent regressions continuously.
|
|
A compiler for deterministic AI workflows. Uses a normative intermediate representation to validate, simulate, and compile portable AI workflows without hiding execution semantics.
|
An open-source infrastructure control plane. Unifies cloud, containers, Kubernetes, and on-prem operations around inventory, approvals, observability, automation, and cost awareness.
|
- Evidence over confidence. A passing claim should point to a check, test, trace, or reproducible artifact.
- Determinism by design. Inputs, decisions, and outputs should remain inspectable and repeatable.
- Security at the boundary. Secrets, untrusted execution, tenant isolation, and destructive actions need explicit controls.
- Honest scope. Planned controls are labeled as planned; simulations are labeled as simulations; limitations stay visible.
- Open and replaceable. Prefer open standards, portable interfaces, and providers that can be changed without rewriting the system.
Build deliberately. Verify independently.
Open to conversations about developer tools, secure automation, and trustworthy AI systems.