Selected work

Useful digital systems, built around a real job.

A selection of live applications, internal prototype work and anonymised workflow systems. Each project is designed to make a specific task clearer, more manageable or easier to review.

How to read these examples

This page shows public projects, internal prototypes and anonymised workflow categories. It is not a collection of client case studies, testimonials, measured results or guarantees.

Agape Duck project area

Live web application

Pawfect Match Finder

A mobile-first dog-adoption experience designed to help people discover and compare dogs looking for homes.

It includes swipe-style discovery, saved dogs, comparison views, compatibility filters, an adopter profile flow, enquiries, bilingual support and administrative dog management.

Evidence boundary: this is a public live application. Features and descriptions are illustrative and may change.

Mobile-first discovery and comparison for people considering dog adoption.

Explore Pawfect Match Finder

Agape Duck project area

Internal prototype

Sky Pilot Simulator

A browser-based drone simulator with two dedicated launch experiences: a portrait touch trainer for phones and a larger flight lab for PC or Mac.

It is an internal educational and research prototype. It is not formal flight training, assessment, certification or operational advice.

Two dedicated interfaces, one shared simulation engine.

Choose a launch experience below

Agape Duck project area

Anonymised workflow category

Intelligent Administration Systems

Source-governed AI workflows that help operators find approved information, draft routine communications, and complete repetitive administration with human review.

These systems can be designed around approved source material, defined boundaries, repeatable tasks and human checking before important outputs are used.

Evidence boundary: this is an anonymised workflow category. No named client, outcome, metric or testimonial is published.

Reviewable support for finding approved information and completing routine administration.

Discuss a similar system

Sky Pilot Simulator

Two dedicated interfaces, one shared simulation engine.

Choose the launch experience that suits your device. Both are educational, experimental prototypes and do not authorise real-world flight.

Best on phone

Portrait · Touch

Sky Pilot Mobile Trainer

A portrait touchscreen drone simulator for learning movement, photography, video and mission awareness using eight focused controls.

Eight controls: U, D, L, R, F, B, P and V.
Missions: Tutorial, Forest and LiDAR.
Compact Flight Review after each run.

Launch Mobile Trainer

Best on PC or Mac

Landscape · Keyboard / Controller

Sky Pilot Desktop Flight Lab

A larger flight and experimentation environment for keyboard or controller operation, programmed routes, telemetry, wind and route optimisation.

Evolutionary Route Lab for programmed and optimised routes.
Wind Experiment for varying conditions.
Telemetry Replay for review and analysis.

Launch Desktop Flight Lab

Both experiences are educational, experimental prototypes. They do not constitute flight training, assessment, certification, or authorisation for real-world flight.

Work areas

What we are building, prototyping and researching.

Status labels are evidence boundaries, not decoration. Live means publicly deployed and currently functioning; working prototype means functioning software or hardware exists while development continues; experiment means a bounded test has been performed; research means active investigation where implementation may be incomplete.

Live + experimental

Evolutionary Drone Simulator

A browser-based drone simulation environment used to investigate route planning, target completion, collision avoidance, return-to-home behaviour and evolutionary search.

Technical themes: evolutionary algorithms, simulation, multi-objective optimisation, route planning, autonomous-systems research.

One recorded experiment: 4,000 genomes and 5,419,023 points were evaluated, producing a simulation run with 3/3 target photographs, zero collisions and 0.6 m HOME error. This is one recorded simulation experiment, not a universal performance claim.

Evidence boundary: results apply to simulation. No evolved controller has flown a physical autonomous drone.

Working prototype / research

Agape Scanner (physical AI)

A custom experimental scanning platform combining a Raspberry Pi 5, RPLIDAR C1, Camera Module 3 and custom 3D-printed hardware.

Technical themes: LiDAR, Raspberry Pi, embedded computing, CAD, additive manufacturing, robotics, digital-twin research.

Evidence boundary: the hardware has been physically assembled and LiDAR Start Scan has succeeded. We are not claiming completed SLAM, sensor fusion, automatic 3D reconstruction or autonomous navigation.

Working capability / product development

3D printing and rapid prototyping

Functional objects, mounts and prototypes developed using a repeated design → manufacture → test → improve process. Real components are measured, parts are designed around them, printed, physically fit-tested and revised.

Consumer-product concepts remain subject to development stage gates and safety and commercial review.

Explore the Development Shelf

Live / MVP / working projects

Custom websites and web applications

Purpose-built websites and lightweight web applications designed around specific user journeys and operational problems, including animal-adoption discovery, drone simulation, charity and nonprofit platforms and operational support tools.

Controlled development practices: GitHub, branches, pull requests, CI, type checking, linting, build validation and responsive testing.

Working prototype / pilot

Volunteer organisation AI

We are using a real volunteer-run sports-club environment to explore how source-grounded, human-in-the-loop AI can reduce administration, preserve organisational knowledge and support operational decisions.

Technical themes: knowledge architecture, source provenance, workflow design, decision support and constraint-aware scheduling research.

Evidence boundary: AI does not run the club and does not automatically schedule a full season. People retain responsibility for decisions.

Have a similar task in mind?

A conversation is a chance to look at the job, the information involved and the review points before deciding whether a website, small app or workflow would be useful.

Start a conversation Start with one practical project How buying works