Mechanical R&D for ocean aquaculture
During an eight-month UVic mechanical engineering co-op with AgriMarine, I contributed to early-stage R&D for a commercial-scale floating closed-containment aquaculture system.
The work took place in a real marine context, where equipment had to be considered around water, biofouling, large composite structures, site access, maintenance, and the practical realities of operating in the ocean.
Designing around a 30-metre floating tank
The main system was a large floating fiberglass tank used for closed-containment fish farming. The tank was roughly 30 metres in diameter and held about 3,000 cubic metres of water.
My work supported practical engineering questions around how equipment would be built, tested, maintained, and improved at that scale.
Testing a cleaner for underwater tank walls
One major task was developing prototype equipment to clean the inside wall of the tank. Biofouling on the tank wall could affect fish health, water movement, drag, and maintenance effort, so the system needed a practical way to remove growth from the submerged surface.
The concept used high-pressure water jets and a spinning disk to help the cleaner stay against the curved tank wall while it moved through the water. I worked on alpha and beta prototype development, testing, and design refinement.
The testing process challenged the original assumptions about how the cleaner was working. Some of the cleaning effect appeared to come from contact and rubbing rather than only from the water jets, which meant the design theory had to be corrected before the prototype could be improved.
Improving the design after the test changed the theory
The next version focused on making the cleaner safer, easier to assemble, and more controlled in how it contacted the tank wall. That included simplifying parts of the assembly, improving the way the device interacted with the surface, and thinking through how the design could eventually become more repeatable and less dependent on manual handling.
I also explored a future semi-autonomous cleaning concept that used the circular flow inside the tank. The idea was to take advantage of the constant water movement, a tether, a spar, and a sail-like element so the cleaner could move around the tank in a more controlled way.
Redesigning the oxygen diffuser for lower-cost manufacture
A second piece of work involved redesigning an oxygen diffuser component. The existing approach relied on a more expensive machined stainless component.
I redesigned it as a waterjet-cut assembly using acetal and stainless plates, with the goal of reducing material and machining cost while keeping the function practical. Based on the estimates available at the time, the redesigned assembly reduced the expected material and machining cost by roughly 58 percent.
Modelling the site and learning from fabrication
Beyond the individual prototypes, I also built SolidWorks models of the site and equipment to help communicate and coordinate the system. That included modelling the tank, barges, bathymetry, pilings, mooring elements, and related equipment layout.
I was also partly involved in early thinking for a newer 30-metre tank and how manufacturing or transport decisions could reduce the cost and complexity of getting large composite structures to site. The co-op also exposed me to composite marine construction through work connected with Blackline Marine and the Quadrant Marine Institute.
Teach a man to fish
This early-career work connected design theory to testing, manufacturing, cost, and environment. A mechanical idea changes once it has to operate in water, be built by real fabricators, be maintained by real people, and survive the constraints of ocean aquaculture.
I still approach design that way: test the assumption, watch what the prototype does, and let the next version respond to reality instead of defending the first theory.
Photo frame guessing game
I designed and built a custom photo-frame guessing game as a personal gift. The idea was to take the familiar logic of a face-guessing game and turn it into something more personal, using real photos of family, friends, and pets instead of generic characters.
The finished piece was built from layered 1/8 inch plywood and laser cut so it could function both as a playable game and as a framed display object when it was not being used. That dual purpose shaped the design. It needed to feel like a finished object worth leaving out, not just a novelty game that would disappear into a closet.
The physical design took more iteration than it first appeared. The flipping photo tiles needed to move smoothly, sit cleanly in the frame, and be easy enough to use without feeling fragile. The plywood layers had to align properly, the tile spacing had to work with the photos, and the matching card set needed somewhere to live so the game stayed self-contained.
I designed integrated storage for the photo cards so the object could hold all of its own pieces. That small detail made the project feel more complete: the game, photo frame, and storage all belonged to the same object rather than becoming separate parts that could be misplaced.
Preparing the photos became its own workflow problem. The source images were inconsistent: different crops, orientations, face sizes, lighting, and backgrounds. To make the game work visually, the faces needed to be prepared in a consistent format and arranged efficiently for printing.
To solve that, I built a small browser-based crop tool. The tool let me crop the source photos into consistent portrait squares and arrange them onto printable 4 by 6 layouts. That made it easier to prepare both the flipping tile images and the matching card set without manually rebuilding every layout by hand.
The project is personal and playful, but it reflects the same kind of thinking I use in more technical work. A simple idea became a product-design problem, a fabrication problem, and a workflow problem at the same time. The final result connected a physical object with a small digital tool so the whole process could be made more consistent, repeatable, and polished.
Parametric espresso tools
After switching to a bottomless portafilter, I ran into two small but persistent workflow problems that did not have good off-the-shelf answers for my setup.
The first problem was cleanup. The spent espresso puck would not always come out cleanly, even with a knock box. I designed a flexible TPU ejector that seals against the bottom of the portafilter basket. A quick smack creates enough air pressure to push the grounds out cleanly into the compost.
The second problem was fit. I wanted a tamping station that actually matched my portafilter, but many printable holders online describe compatibility by brand or general size without exposing the dimensions that matter. In practice, the fit depends on details like basket diameter, ear geometry, handle angle, saddle height, and counter clearance.
To make the design more adaptable, I created a parametric OpenSCAD tamper station. Instead of remodelling the holder for every variation, the key dimensions can be measured, entered, and regenerated for a specific portafilter setup.
This is a small project, but it captures a useful design pattern: solve the immediate problem, identify which dimensions actually control fit and function, and turn the solution into something adjustable enough to be reused instead of rebuilt from scratch.
Fridge door closing magnet
My fridge door would sometimes stop just short of closing fully. It was a small problem, but the result was annoying: the door could look almost closed while the seal had not fully taken over.
I considered a few possible fixes, including tipping the fridge back slightly, adding an alarm, or designing a retracting magnet mechanism. Those options would have added complexity or solved the problem after it happened rather than helping the door close properly in the first place.
The simpler answer came from looking more carefully at the fridge itself. I noticed an unused feature from the reversible-door design: a spot where the door hardware could have been mounted if the hinge direction were changed. That gave me a clean place to attach a small helper without modifying the main door or cabinet.
I designed and 3D printed a small holder for rare earth magnets. The magnets provide just enough pull to bring the door through the last bit of travel so the edge gasket can take over and seal the door normally.
It is a small project, but it captures a useful design pattern: look at the actual object, notice the features that are already available, and choose the simplest physical intervention that solves the real failure mode.
Fridge circulation fan
My Danby all-fridge was freezing food on the lower shelf while the rest of the cabinet stayed at a normal temperature. The drain was clear, and the fridge appeared to be operating normally, so the problem was not obvious from a standard check.
I used a thermal camera to look at what was actually happening inside the cabinet. The images showed cold air pooling near the lower back of the fridge. The unit relied mostly on natural convection rather than a built-in circulation fan, so the coldest air was not mixing evenly through the cabinet.
The fix was a small targeted airflow improvement rather than a major appliance modification. I added a low-voltage fan on a custom 3D-printed bracket to gently circulate air through the cabinet. I also added a small thermal switch attached to the compressor so the fan only runs when the fridge is actively cooling.
The result was a more even temperature distribution and a practical fix for a frustrating problem. It is a small project, but it shows a useful pattern: diagnose the actual system behaviour, avoid unnecessary complexity, and make a targeted physical improvement that matches the real failure mode.
And now I have no excuse not to eat my vegetables.
