Skip to content

Infrastructure & Embedded Systems Engineer

Designs and operates ESP32-based Zigbee systems and production-style infrastructure under real-world constraints.

View Infrastructure See Projects

At a Glance

  • Designed and built 2 custom ESP32 Zigbee devices from scratch — deployed to a live environment and validated under continuous operation
  • Custom firmware operates within a larger Home Assistant environment alongside commercial devices — the live deployment is the test environment
  • Implemented and operate a KVM/libvirt platform hosting 10+ VMs on bare-metal hardware, running continuous production services
  • Designed a VLAN-segmented network with pfSense, 10GbE primary path, bonded interfaces, and automatic 1GbE failover
  • ZFS-backed storage supporting VM workloads, with snapshots, replication, and integrity guarantees
  • Built GitHub Actions pipelines that automate firmware builds, OTA image packaging, and versioned release delivery

Systems I Currently Operate

These systems support daily workloads and have evolved through hardware upgrades, network redesigns, and service migrations over several years.

Compute & Virtualization

Built on a Dell PowerEdge T620 (dual Xeon) running KVM/libvirt. 10+ VMs host continuous services; hardware passthrough is configured for dedicated network functions and media transcoding. Stability-critical workloads are isolated to separate guests to contain failure blast radius.

Storage

Implemented on ZFS for filesystem-level integrity guarantees. Snapshot and replication workflows run on schedule between systems. Health is tracked through monitoring and regular scrubs; recovery always starts from a known-good snapshot.

Network

Designed around segmentation first: IoT, infrastructure, and trusted zones are separated at the VLAN level with pfSense enforcing policy at every boundary. Primary connectivity is 10GbE with bonded interfaces; a 1GbE path takes over automatically on link failure with no manual intervention required.

Automation & Embedded Platform

Home Assistant coordinates a mixed environment of commercial and custom devices. Custom Zigbee devices (ESP32-H2/C6 firmware, built here) run alongside off-the-shelf hardware — integration with a real, heterogeneous environment drives design decisions. The Zigbee mesh uses a Texas Instruments CC2652 coordinator. Self-hosted services include network monitoring, a browser-based editor, and infrastructure tooling.

Monitoring & Delivery

Service availability, hardware health, storage status, and replication success are tracked continuously. GitHub Actions pipelines automate firmware builds, OTA image packaging, and release publishing.

Featured Work

LD2450 Presence Sensor
ESP32-H2/C6 firmware for a 24GHz mmWave radar — built from scratch because no native Zigbee implementation existed. Replaces binary presence detection with configurable polygon zones mapped to physical space. Coordinator fallback keeps lights operating autonomously when Home Assistant fails; a two-tier watchdog distinguishes transient jitter from genuine outages, preventing false automation triggers. Addresses reliability issues in Zigbee presence detection, including missed updates and automation delays.
Zigbee LED Controller
ESP32-H2/C6 firmware driving two LED strips through 8 independent Zigbee endpoints. SPI replaces RMT to resolve a hardware conflict with the Zigbee radio. A 200Hz firmware-side transition engine compensates for the coarse interpolation resolution of the Zigbee stack.
Network Architecture
Segmentation-first design: IoT, infrastructure, and trusted zones isolated at the VLAN level. pfSense enforces policy at every boundary. 10GbE primary path with bonded interfaces; automatic 1GbE failover validated under real link-loss conditions.

Engineering Approach

Design for operability

  • Redundancy where failure impact is high; simplicity elsewhere
  • Detect failures through instrumentation before users do
  • Validate changes before permanent adoption; maintain rollback capability

Root cause over quick fix

  • Investigate failures rather than work around them
  • Balance complexity against operational risk
  • Isolate workloads to limit blast radius

Grounded in real failure

  • Disk failures, hardware troubleshooting, driver incompatibilities
  • Network failover testing and replication anomalies
  • Infrastructure decisions shaped by actual incidents, not hypotheticals

What You’ll Find Here

  • Infrastructure: Network design, storage strategy, compute platform, and operational practices
  • Projects: Embedded Zigbee devices — design decisions, implementation challenges, production deployment
  • Skills: Technology competencies with supporting evidence

This site showcases infrastructure and embedded systems engineering work.