I'm contracting at Pixomondo, part of Sony, as a senior iOS and visionOS engineer. I joined in May 2025.
The main thing I work on is a real-time AR video capture app for iPhone, iPad and Apple Vision Pro. It lets a VFX supervisor judge how CG assets sit in a live shot while the crew is still on set, rather than finding out once it reaches the edit. That feedback loop used to run for days. Now it runs while everyone is still standing there.
My part of it is the rendering. Custom RealityKit entity shaders and full-screen Metal post processing — outline passes, and emulation of cinema cameras including the Sony Venice — holding 60fps on device. The point of the camera emulation is that a supervisor can preview a look without anyone rigging the physical camera to match.
Separately, I sole-authored the AI generation pipeline behind a Harry Potter location-based experience. A guest uploads a photo and gets back a personalised video of themselves in the wizarding world, generated against supplied reference imagery. It's serverless — Step Functions, Lambda and DynamoDB, provisioned with Terraform — and it has served thousands of guest generations.
The less visible work is the part I'd point at if you asked what changed. I migrated a four-engineer iOS team from GitLab to GitHub, and from GitFlow to GitHub Flow, and replaced a two-hour manual TestFlight release with an automated build on every commit to develop. Release cadence went from ad-hoc to continuous, and the team got those two hours back.
I also wrote the automated PR checks the engineering team now uses — duplication detection and diff-size limits. They exist to keep pull requests small, because small pull requests get reviewed and large ones sit there.