I build API integrations, inspectable automation, and developer tools that explain what happened, why it failed, and how to recover.
My portfolio focuses on practical engineering questions: validating a tool call before execution, preserving webhook state under retries, making API errors actionable, and keeping human review visible.
Explore my portfolio · Recruiter review guide · LinkedIn
- API and developer experience: inspect DX Orbit, an OpenAPI scorecard with traceable findings, revision comparison, and report export.
- Agent tool boundaries: inspect ToolTrust, a deterministic lab for JSON Schema validation, permissions, human confirmation, and simulated execution traces.
- Full-stack workflow engineering: inspect ForgeFlow, including validated inputs, persisted run history, visible fallback behavior, and human review.
Open the tests and CI alongside the source. They show the behavior each project is designed to preserve.
| Project | Engineering question | Evidence to inspect |
|---|---|---|
| DX Orbit | Where does an API contract create integration friction? | Explainable scoring, JSON/YAML input, malformed-input checks, comparison reports |
| ToolTrust | Should this tool call proceed? | Schema validation, permission and confirmation gates, readable replay traces |
| DevStart | Can a developer recover from an API failure? | Six deterministic onboarding scenarios, recovery guidance, session benchmarks |
| HookForge | Does state survive duplicate and out-of-order delivery? | HMAC checks, duplicate protection, stale-event rejection, bounded simulations |
| SignalDesk | Which documentation problem should a maintainer investigate next? | Feedback taxonomy, weighted priorities, resolution workflow, issue drafts |
These are runnable engineering labs with explicit limits. Simulated delivery and tool execution are labelled; fictional feedback is not customer research. They do not claim production adoption or measured business impact.
- MailTrace DX Lab: signed webhooks, retries, idempotency, event ordering, and developer-facing debugging.
- ForgeFlow AI Automation: React/TypeScript, FastAPI, SQLite persistence, testing, Docker, and CI.
- ClearRoute API: typed validation, explicit state transitions, predictable error contracts, and audit events.
- AccessPath Console: keyboard-first interaction, visible focus, recovery states, and accessibility checks.
- Technical Portfolio / Signal Engine: an interactive review path connecting the product experience to source, tests, and implementation boundaries.
- Make inputs, state transitions, and failure paths explicit.
- Validate before acting; keep permissions and human decisions inspectable.
- Test the behavior that matters, including recovery and malformed requests.
- Document what runs today and what production deployment would still require.
Working stack: Python · FastAPI · Pydantic · JSON Schema · REST/OpenAPI · TypeScript · React · SQLite · pytest · Vitest · Docker · GitHub Actions
Role focus: AI automation, technical solutions, API integration, developer experience, and implementation engineering.