Roderick McLeod

A kinetic launch startup building a ground-based accelerator. I was handed structural problems with short fuses and expected to carry them from requirement to finished hardware.

• Designed and manufactured the rolling stand for the Accelerator Demonstrator, an 11 foot diameter 3/4 inch thick 20000lb tube inside my first 2 weeks at Longshot.×
• Had had Chose an I-beam structure on stiffness per dollar, boxed the sections that needed torsional rigidity, and sourced to what suppliers actually had on the shelf rather than to an ideal BOM.×
• Sized the swivel-caster mounts with FEA in PrePoMax, then checked the bending moments by hand before anything was cut. Neither method gets trusted alone.
• Got the plasma cutter working on day one, diagnosed and repaired its limit switches, chose the CAM software, developed cut parameters for ¼-inch steel and ⅛- and ¼-inch aluminum, then trained others to run it.×
• Built mounting hardware for test campaigns and ran them. Caught a structural failure mode during testing that was corrected in the next revision.
• Created the "Investor Projectile" it is a projectile that was given to the first 100 Longshot Investors, constraints were for it to survive being shot, be under $200 per projectile, and would need to be assembled in under 1 hour. I managed to get that down to $125 per projectile and 20 minutes for assembly, I later found out it was tested, and that the parts had been ordered with no changes made. Order total: $12500. ×

The stand, on LinkedIn →

• structural design
• FEA / PrePoMax
• design for manufacture
• TIG / MIG welding
• CNC plasma
• test campaigns

Private note, not public: Cleared as public: the stand, the demonstrator tube on the cart, and the projectile. Testing stays generic — no architecture, no campaign specifics, and the instrumented test hardware photos stay out.

A competition to build the structurally most efficient I-beam: carry a design load over a 23-inch span in three-point bending, at the lowest possible mass. I ran the full build-test-iterate cycle, and welded the tooling that made the builds repeatable enough to iterate on at all.

• Took 1st and 2nd in specific strength in Category A at SAMPE 2025 — carbon-fiber I-beams, judged on load carried per gram.
• Built the lightest beam in the field to meet design load: 431 g carrying 7,200 lb.
• Designed and TIG-welded the steel behind it — manufacturing jigs, alignment clamps, test fixtures. The V2 aligner cut build-to-build variance enough that layup changes became measurable instead of noise.
• Refined layup schedules and fabrication methods across iterations rather than copying a known-good recipe, which is why the second entry beat the first.
• On the natural-fiber side, found that a bamboo and spruce flange-web architecture outperformed more complex flax-composite approaches while being markedly simpler to build — those beams placed 2nd and 3rd lightest to reach design load.
• Founded the SJSU chapter, trained its officers, and guest-lectured at San José State and Cal Poly SLO on beam design and competition strategy.

• prepreg composites
• vacuum bagging
• TIG welding
• jig design
• destructive testing
• failure analysis

A tilt-trirotor that lifts off vertically on three rotors, then tilts its 22-inch forward pair to transition into wing-borne cruise. I own the aerodynamic characterization and the control-system model.

• Built a 1.02 M cell polyhedral mesh holding y+ under 31 across the full wing — down from 163 on the first attempt, after rebuilding the CAD to kill sliver triangles and a knife-edge trailing edge.
• Calibrated the transition model against UIUC wind-tunnel polars, then swept angle of attack to find peak L/D of 18.95 at 3° — which lands on the cruise condition rather than near it.
• Checked the result against Prandtl lifting-line theory independently: span efficiency e = 0.979, within 4% of the elliptic closed form.
• Settled the winglet argument with numbers. Induced drag is 40.6% of the total, but only 0.9% is actually recoverable — and a span extension of equal added area beat winglets 3.8% to 1.9%.
• Deriving a state-space plant that carries structural flex as modal coordinates, to compare LQR against PID-with-notch on the same model.
• Coupling Simulink to ArduPilot over the SIM_JSON interface at 400 Hz — my aerodynamics as the plant, the real firmware as the controller.

• OpenFOAM
• ANSYS Fluent Meshing
• SolidWorks
• Python
• MATLAB / Simulink
• LQR
• ArduPilot

A planner for an engineering course load, built because the generic ones kept lying to me about how much time I had. One self-contained page, roughly 3,300 lines.

• Pulls the assignment manifest from Canvas and reconciles it against the syllabus — one of my courses posts graded work only in the syllabus table and never in the assignments tab, so a naive scrape silently misses real deadlines.
• Models late-penalty policy per course, so it can tell a deadline that costs 10% from one that costs the whole grade, and prioritize accordingly.
• Lays the day out against fixed commitments — class, shop time, training — so the plan reflects hours actually available rather than hours nominally free. That distinction is the entire point.
• Flags work that is blocked rather than merely unstarted, which is the failure mode that actually wrecks a week.
• Worth building because a planner this specific — one person, one course load — would never have justified the hours before. That calculation has changed.

Open the demo →

• single-page app
• Canvas LMS
• scheduling
• vanilla JS

Private note, not public: The live planner holds my real schedule and stays private. The linked demo is the same interface with entirely fabricated courses and deadlines.

A ground-up electric scooter — pack, drivetrain and chassis hardware. The kind of project where the electrical and the mechanical failure modes are both yours to find.

• Designed and assembled a custom high-current battery pack: cell selection, pack layout, balancing, BMS integration, and wiring sized to match.
• Fabricated the steel motor mounts and drive hardware, including belt and chain components.
• Tuned the BLDC motor and ESC as a system rather than to spec-sheet numbers.

• battery pack design
• BMS
• BLDC / ESC tuning
• steel fabrication