Software Engineer · Python · Test Infrastructure · Backend · AI Applications
Software engineer with an electrical engineering and aerospace/defense background. I spent eight years across hardware validation, systems integration, qualification, manufacturing systems, and failure investigation, where I increasingly focused on Python automation, telemetry, debugging, test infrastructure, software tooling, and hardware/software integration. I now build Python backend systems, developer/internal tools, automated test infrastructure, and AI applications with an emphasis on reliability and measurable verification.
Target roles: Software Engineer · Test Infrastructure / Automation · Python Backend Engineer · Developer / Internal Tools · AI Application Engineer
| Area | Engineering focus |
|---|---|
| Test infrastructure & automation | Python test frameworks, hardware/software integration, CI/CD, failure isolation, repeatable validation, and developer tooling |
| Backend systems | FastAPI services, typed contracts, SQL persistence, asynchronous work, health checks, and failure handling |
| Internal tools & automation | Engineering workflow automation, API integrations, operator tooling, debugging, and repeatable verification |
| AI applications & infrastructure | Local vLLM inference, request admission, context budgeting, tool/agent workflows, performance measurement, and rollback controls |
| Telemetry and hardware integration | Binary protocols, CRC validation, serial communication, device detection, and evidence capture |
Five-service Python system that receives a fixed binary telemetry frame, preserves the original bytes, validates the interface contract, coordinates durable processing through PostgreSQL and Redis, and returns operator-readable results.
Verified behavior: 146 automated tests · isolated end-to-end verification · raw-byte preservation · idempotent duplicate handling · sequence checks · CRC rejection
Demonstrates: service boundaries, durable state, API design, binary protocol handling, failure-path testing, and Docker Compose integration.
Fail-closed Python workflow for serial and CAN hardware validation. It detects Linux devices, records exact command/response evidence, and prevents software-only checks from being presented as physical hardware validation.
Demonstrates: PySerial, hardware presence gates, structured evidence artifacts, explicit failure behavior, and the boundary between simulation and observed bench results.
These repositories remain private because they contain workstation-specific operating configuration rather than public portfolio material:
- AIWork control plane — executes bounded agent changes in isolated Git worktrees, verifies them independently, requires human ownership attestation, and rolls back failed applications.
- Local AI runtime — versions the vLLM container configuration, health checks, measured routing experiments, and rollback profiles for a local NVIDIA GPU inference stack.
- Qwen request governor — provides OpenAI-compatible request sanitation, context budgeting, admission control, upstream failure handling, and request audit records.
Python · FastAPI · Pydantic · SQLAlchemy · PostgreSQL · Redis · pytest · Docker Compose · Linux · systemd · vLLM · GitHub Actions · serial/CAN telemetry
Code existence is not proof of system behavior. Claims are limited to behavior that was directly exercised, checked against explicit acceptance criteria, and preserved as repeatable evidence.