Happy Europe Day! 👋 It’s been an interesting year so far, as we see a significant push to building in sovereign European data centers. We’re working with a number of sustainability-focused compute providers to make building in Europe as easy as building in the US, (and more cost-effective).

🧩 Privacy-Preservation with Smarter Automotive Experiences by @fillip.pro @ Wasm I/O 2025 ▶️ Video: youtu.be/rh9De686BSM 🔗 Slides: 2025.wasm.io/slides/priva... #wasmio25

Exploring the missing piece of AI with WebAssembly at #Wasmio25: training. Around 6-years of server-side WebAssembly, and WebNN (privacy) work has led me back to the edge. We’ve come along way and I’m amazed at what the community has built, (and which I’ve had the benefit of building on).

Attending GTC this week where Nvidia, Jaguar Landrover, Zenseact, and Volvo are demoing robotics and self-driving cars. I’m showcasing how to build safer AI solutions by running training on-device, locally in a secure PDS, whilst collecting zero data. There’s no better time for safety-by-design.

Attending the Nvidia GTC event this week where we’re focused on the blueprints for building smarter and more sustainable in-cabin experiences for EVs. We’re using federated learning, digital twins, and a lot of WebAssembly to orchestrate privacy-preserving intelligence training workloads.

If you or someone you know is an awesome rust dev 🦀 who’d be interested in working together (part-time) for the next 6 months to build out rsky - please DM me. Familiarity with atproto and especially the rsky codebase a huge plus.

In the long term there are zero reasons to be sharing data abroad. In the near future we need to be rearchitecting approaches to data usage to reduce the dependency on data sharing for improving services and research capabilities. Privacy enhancing technologies are largely being underutilized.

This project uses an in-development Naamio Space runtime, which stores data in a Secure Enclave and enables it to be used for training and inference of AI models on-device using WebAssembly and gRPC for privacy-first polyglot AI.