Custom-built drone platforms for heavy lift, waterproof, and Antarctic research operations.
1 - LoRaScope
A single-radio LoRa / MeshCore / LoRaWAN spectrum and traffic analyser I designed and build.
LoRaScope — see what's actually on the air
A single-radio analyser that listens to the LoRa band and tells you what's really happening on it — live capture, packet decode, and link health in one tool.
LoRaScope is a product I designed, build, and consult around. It captures LoRa
traffic live, decodes packets, and turns raw airtime into a clear picture of
your spectrum — who’s talking, how often, and whether the link is healthy. It
works at the radio layer directly, so you can debug a site even where there is
no LoRaWAN network server and no gateway to log into.
One analyser covers every flavour of LoRa — LoRaWAN, LoRaBASIC, MeshCore,
and more — all through the same interface.
What it does
Capture & decode
Match a sender’s sync word, spreading factor, bandwidth, and coding rate, and
LoRaScope decodes raw frames the moment they hit the radio — no gateway, no
join, no network server. Save them to a capture store to filter, search,
export, and replay later.
Spectrum & link health
Turn live airtime into a readable picture — RSSI, SNR, source, payload,
collisions, and duty cycle — so problems on a busy or marginal band become
obvious instead of guessed at.
Emulate & inject
Pretend to be the hub, or any node. Inject and emulate traffic to test a node
in isolation, reproduce a fault, or validate firmware — without disrupting
production hardware.
Agent & script control
The live feed and the controls are exposed over telnet and REST, so a
script — or an AI agent — can monitor, reason, act, and observe in a loop. See
AI & Automation.
Related work
LoRa Gateway — bridging a local LoRa network to the wider world
A local LoRa band can be busy — dozens of sensors reporting temperatures, tank
levels, gate states, and soil moisture every few minutes. You rarely want all
of that going out over an expensive satellite link. The gateway watches every
packet and, by rules you define, forwards only what you asked for — a daily
aggregate over Iridium at 09:00, an SMS if a pump fails, a live mobile-data
dashboard for the operator on the hill.
What it does
Multi-transport backhaul
Forward traffic over LoRaWAN to a remote site, up to Iridium / LEO satellite
when there is no tower in sight, across LTE mobile data, or as plain
SMS alerts — queuing and transmitting as each link allows.
Rules engine
Define what leaves the site and when — thresholds, schedules, aggregates, and
alerts — so costly links carry only the data that earns its airtime.
MeshCore relay
Bridge a local network into a wider MeshCore mesh for coverage around obstacles
and across distance, with multi-radio setups where needed.
Remote access tunnel
Reach the gateway and the devices behind it from anywhere, securely, for
configuration and diagnostics — no truck roll to a remote site.
Related work
LoRaScope — the analyser the gateway is built alongside
Instrumenting LoRa networks so an AI agent — or any script — can monitor, test, and diagnose them in a closed loop.
Give an AI agent eyes and hands on the LoRa band
Most analysers are a screen a human stares at. I build the tooling that lets an agent — or any script — watch the band, act on it, and observe the result, autonomously.
LoRaScope exposes both its live feed and its
controls over telnet and REST. An agent can watch the band, read decoded
packets, run custom decoders, then inject and emulate traffic to test a
hypothesis and watch the result — monitor, reason, act, observe, in a loop.
Because it works at the radio layer, an agent can instrument a site even where
there is no LoRaWAN network server and no gateway to log into.
What it enables
See — agent-readable
Stream the live decoded-packet feed over telnet or REST — RSSI, SNR, source,
payload, timestamps — and query the capture store, link health, and historical
logs. Every packet is also logged to SD as human-readable CSV.
Act — agent-controllable
Switch channel and protocol, inject packets, and emulate a hub or node through
the same single API — one call, no separate tool per stack (LoRaWAN,
LoRaBASIC, MeshCore, and more).
Loop — closed-loop diagnosis
Combine the two and an agent can run autonomous fault diagnosis: form a
hypothesis, inject a test, read the result, and narrow down a problem without a
human on site.
MeshCore, LoRaWAN and ChirpStack work — the radio and networking layer beneath my monitoring products.
Mesh & LoRa Networking
The radio and networking layer that ties sensors, hubs, and gateways together across distance and obstacles.
Overview
Most of my remote-monitoring work runs on long-range, low-power radio rather
than telco. I design and deploy across the full LoRa family — point-to-point
LoRa, MeshCore mesh networks, and LoRaWAN with a ChirpStack network
server — picking the right stack for the site rather than forcing one pattern
onto every job.
What I work with
MeshCore
Self-healing mesh for coverage around obstacles and across distance. I run the
AU915 Victoria preset by default, with an SF12 / BW125 long-range alternative
for the hardest links, across Companion, Repeater, Sensor, and Room Server
firmware roles.
LoRaWAN & ChirpStack
Standards-based LoRaWAN with a self-hosted ChirpStack network server, Cayenne
LPP payload encoding, and integration into local and cloud data pipelines.
Hardware
ESP32, nRF52840, RAK4631, Heltec V3, SenseCAP T1000-E, and RP2040-class nodes —
custom firmware in C++ with ultra-low-power deep-sleep designs for multi-year
battery life.
Tuning & diagnosis
When the defaults do not fit a site, I tune the link and prove it with
LoRaScope — measuring spectrum, collisions, and
link health at the radio layer rather than guessing.
OGLAS — custom sensors, hubs, and data analysis for off-grid farms, fleets, factories, and remote sites.
OGLAS — your site, your data, your alerts
The Off-Grid Local Alert System — custom sensors, hubs, and data analysis for anywhere there's no reliable phone line and no appetite for a monthly cloud bill.
OGLAS is the monitoring product line I build for off-grid and poorly-connected
sites — farms, fishing boats, golf-cart fleets, forklifts, factories, and remote
power installations. It measures and controls things in the field, gets the data
to where you can use it, and tells you when something needs attention.
The pipeline is simple — Sensors → Hubs → Local → Cloud — and
I build the custom version of any link in that chain.
No subscriptions. No telco dependency. Your data is your data.
How it fits together
Sensors
Measure or switch anything — tank and trough level, soil moisture, power, gate
and asset state, vehicle and engine, weather, smoke and gas. Long-range
wireless, built to run off-grid.
Hubs
On-site receivers that collect sensor data and hand it on — LoRa, mesh, and
satellite hubs, with optional large or small displays. See
Mesh & LoRa and the
Gateway.
Local
Browse, analyse, and own your data on site — on an SBC or a laptop. Backs up and
manages data across every hub. It stays yours, and it stays local.
Cloud
An optional private copy off-site — over LTE, Wi-Fi, satellite, or HaLow. View
it from your phone while you’re away, and share it with whoever you choose.
Custom builds
The real work is bespoke: sensors, hubs, monitoring, alerts, and data analysis
and presentation for sites where nothing off the shelf fits. If you have a thing
that needs watching — online or off-grid — that is exactly what I do.
Production Zephyr firmware — BLE provisioning, WiFi, LoRa, OTA, crash reporting, storage, displays, and multi-year battery life, proven in the OGLAS v2 platform.
Zephyr RTOS — production firmware, not tutorial code
The full connected-device stack on Zephyr — BLE configuration,
WiFi, LoRa, over-the-air updates, crash reporting, storage, displays, and
multi-year battery life — shipping today in the second-generation OGLAS
firmware across nRF52840, ESP32, and STM32.
Zephyr is my RTOS of choice for serious connected products. It is
vendor-neutral — the same codebase runs on Nordic, Espressif, and ST silicon
— with devicetree and Kconfig keeping board differences out of application
code, west managing builds and dependencies, and first-class subsystems for
Bluetooth, networking, power management, storage, and firmware update.
These are not isolated demos. Every capability below is shipping in the
second-generation OGLAS firmware —
hubs and sensors deployed to off-grid sites where a truck roll to fix a
locked-up device is not an option. That constraint drives everything: devices
must configure themselves in the field, update themselves remotely, report
their own crashes, and run for years on a battery.
Capabilities
BLE configuration & provisioning
Custom GATT services for on-site commissioning — set WiFi credentials, radio
settings, and site parameters from a phone, with secure pairing. No serial
cable, no reflash, no laptop in a paddock.
WiFi
Station and access-point provisioning flows, credential storage, and robust
reconnect handling for networks that come and go — including BLE/WiFi
coexistence on ESP32-class parts.
LoRa & long-range radio
Zephyr LoRa on SX126x-class radios (RAK4631 / nRF52840) — the radio layer
under the OGLAS sensor network, backed by the
Mesh & LoRa and
Gateway work.
Long-term battery management
Zephyr device power management, sleep states, and duty-cycling for multi-year
field life — plus battery telemetry so the data tells you a device is fading
long before it goes quiet. Ultra-low-power deep-sleep design is a speciality.
OTA updates
MCUboot dual-slot images with signed firmware and safe rollback — updates
delivered over BLE or WiFi to devices you may never physically touch again.
A failed update falls back; it never bricks.
Crash reporting
Fault handlers, coredumps, and persisted crash and reboot-reason logging, so
a field failure comes home with the data instead of vanishing into a reset.
Rare bugs get found and fixed, not shrugged at.
Storage & settings
NVS and the Zephyr settings subsystem for configuration and calibration, and
flash filesystems for data logging — all designed to survive power loss,
because off-grid power fails without warning.
Displays
OLED, eInk, and LCD panels under Zephyr’s display subsystem and LVGL — from
always-on status screens to ultra-low-power eInk that costs nothing between
refreshes. See Custom Displays.
RTC & timekeeping
Battery-backed RTC and time synchronisation across sensors and hubs, so data
from an off-grid site is correctly timestamped even after weeks with no
uplink and no NTP.
Platforms
One application codebase, per-board devicetree overlays and Kconfig
fragments — that is the point of Zephyr, and it is how the OGLAS v2 firmware
is structured.
Platform
Boards
Strengths
nRF52840
RAK4631, SenseCAP T1000-E
BLE + SX1262 LoRa, exceptional sleep currents
ESP32 / ESP32-S3
Heltec V3 and custom boards
WiFi + BLE, displays, plenty of headroom
STM32
Custom and off-the-shelf boards
Wide part range, industrial peripherals, RTC
Proven in OGLAS v2
The second-generation OGLAS hub and
sensor firmware is built on Zephyr and exercises everything on this page in
production: BLE-commissioned, LoRa-connected, OTA-updatable devices that log
locally, report their own faults, and run unattended for years. If your
product needs the same, this is exactly what I do.
Purpose-built APIs with full documentation, authentication, load control, and testing tools.
What I Build
Features
Full documentation — auto-generated, always up to date
Authentication — token-based, role-based, or custom schemes
Load control — rate limiting, queuing, and graceful degradation
Testing tools — built-in test interfaces for development and QA
Versioning — backward-compatible API evolution
Approach
REST-first design with clear, consistent endpoints
Designed for both internal integration and external consumption
Comprehensive error handling and status codes
Performance-optimised for high-throughput environments
Built to integrate with existing systems and data sources
Documentation & Testing
Built-in API documentation and interactive testing interface
8 - Custom Authentication
Custom Authentication Solutions
Unified identity management across multiple sites with flexible authentication methods, admin dashboards, and database views.
Capabilities
Multi-Site Identity
Single authentication system spanning multiple web properties with unified user management and consistent login experience.
Admin Dashboards
Custom management interfaces for user administration, role assignment, and account oversight — built for the people who actually use them.
Flexible Methods
Support for multiple authentication flows — token-based, session-based, or custom schemes — with role-based access control throughout.
Screenshots
User management — view and edit user details, roles, and permissions
Custom database views and tables for direct data inspection
9 - Custom Displays
Embedded Display Solutions
Purpose-built display systems for solar monitoring, industrial data visualisation, and long-running embedded applications.
Capabilities
Any Size
From tiny status indicators to full OLED and eInk panels — matched to your use case and environment.
Always On
Designed for long-running, unattended operation with automatic logging and ultra-low power consumption.
Interactive
Touch input and scroll navigation on embedded displays — real controls, not just a read-only screen.
Featured Build — Battery E-Paper Dashboard
A battery-powered colour e-paper dashboard — site readings, battery states, and a calendar, running for months on a charge
A wall-or-bench dashboard on 800×480 colour e-paper (E Ink Spectra 6, ESP32-S3). It renders typed pages — a home summary, calendar, power, battery — each tile stamped with its data age, and the whole unit is provisioned from a phone: an unconfigured panel comes up as its own access point with a QR code to join it.
E-paper costs nothing between refreshes, so at a 90-minute update cadence the panel runs about three months per charge — the display budget becomes how often you draw, not how fast.
Resilient Networking
Knows up to four WiFi networks and joins the strongest in range, with primary and fallback REST data endpoints and on-screen health indication — the footer shows which source drew the screen.
Months Per Charge
The picture stays on the glass with the power off. Wake, fetch, draw, sleep — around three months on a cell at 90-minute refreshes, with its own battery state in the footer.
Sensors Onboard
Reports its own temperature, humidity, and battery voltage alongside the data it displays — the panel is a sensor node in the room as well as a window into the site.
Solar monitoring on a tiny display — compact, low-power, always visible
Solar monitoring on a medium display with touch input and scroll navigation
10 - Dashboards
Custom Dashboards
Real-time monitoring, firmware management, and device configuration — built to run on anything from a Raspberry Pi to a full server.
What Makes These Different
Local data, local control — dashboards connect directly to your hardware and data sources
Any size, any type — from compact embedded displays to full-screen monitoring walls
Long-running reliability — designed for always-on operation with automatic logging
Low power options — OLED and eInk display support for battery-powered deployments
Examples
Firmware upload and management
Real-time monitoring
Device settings and configuration
11 - Drones
Custom Drone Platforms
Purpose-built drones for research, industrial, and remote environment applications — including Antarctic deployments.
Overview
Over the years I have designed, built, and operated custom drone platforms for a range of demanding environments. These builds predate the current generation of commercial drones and were purpose-built to meet specific mission requirements.
Heavy Lift
Platforms capable of lifting payloads up to 20 kg — built for carrying research instruments, cameras, and sampling equipment in the field.
Waterproof
Designed for operation in rain, spray, and wet Antarctic conditions — sealed electronics and weather-resistant airframes.
Antarctic Operations
Flown in harsh polar environments for research support in 2014 — extreme cold, high winds, and remote operation.
Local Operations
Aerial photography, surveying, and testing in Victoria, Australia — proving ground for Antarctic-bound platforms.
Builds
Photos
12 - Embedded Dashboards
Single-File Embedded Dashboards
Zero-dependency, self-contained HTML dashboards designed to run on resource-constrained devices like the ESP32.
How It Works
A single HTML file containing all SVG, CSS, and JavaScript needed to control and monitor your embedded device. No external dependencies, no CDN, no internet connection required.
Example Features:
Light controller with auto-discovery
WiFi search and configuration
Full device status and memory monitoring
Multiple chart types for data visualisation
Also Built In:
File management interface
JSON API integration
Auto-discover configurable settings
Responsive design for any screen size
Screenshots
Light controller — auto-discovery checkboxes and status
Full device status including free memory
WiFi search and configuration
Multiple chart types available
13 - WII
Waves In Ice
Custom wave monitoring systems designed for Antarctic sea ice — deployed internationally and published in Nature.
Research by the National Institute of Water and Atmospheric Research (NIWA) in New Zealand has revealed that large ocean waves, particularly those over 3 metres, can travel hundreds of kilometres into Antarctic sea ice, acting as a major force in breaking up ice floes and accelerating retreat. This phenomenon explains how storm swells in the Southern Ocean significantly impact the stability and extent of polar ice.
Key Findings
Significant Impact Range: Large waves from the Southern Ocean can penetrate and break sea ice up to 350–500 km from the ice edge.
The “3-Metre” Threshold: Waves smaller than 3 metres generally have their energy dissipated quickly, but waves larger than 3 metres create a sustained, destructive oscillation in the ice pack.
Mechanism of Breakup: As waves pass through the Marginal Ice Zone (MIZ), they break the ice into smaller pieces. These smaller floes are then more easily deformed and broken further by winds and currents.
Impact on Ice Extent: Increased storminess and larger wave heights in the Southern Ocean are linked to increased breakup and retreat of sea ice, particularly near the Antarctic Peninsula.
Role in Climate Change: This research bridges the gap in understanding why some areas of sea ice grow while others recede, highlighting that wave-driven ice fracturing is a key, previously underestimated factor in climate models.
Innovative Methodology: Due to harsh conditions, earlier studies were limited. NIWA scientists used specialised, robust, autonomous equipment to measure the effects of these waves directly from the ice, including during and immediately after large storm events.
This research, largely published in Nature in 2014, fundamentally changed the understanding of how ocean dynamics affect the longevity of sea ice.
