Founding Machine Learning Engineer

Founding Machine Learning Engineer About the job

Who we are

Voyage is a Maison for Mind Computing – the first deep tech couture house, engineering non-invasive neural interfaces as objects of desire. With a founding team from Apple, MIT Media Lab, Harvard, the Royal College of Art and more, we blend frontier science with cultural gravity to create technology that is equally functional, aesthetically refined, and highly adoptable. All team members are known as ‘Technical Philosophers’ and collaborate at the intersection of hardware, AI/ML, neuro, design, and philosophy of mind.

Applicants can expect a creative, experimental environment focused on pushing the boundaries of human-machine interaction.

Most ML roles begin with “which architecture fits this data.” We like to start with “what is this data, actually.”

There’s no established benchmark for this, no textbook preprocessing pipeline, no prior work to sanity-check yourself against. You’re not fitting a model to a known modality – you’re the first person deciding what the modality even is, both structurally (what’s actually sitting in the raw signal) and functionally (how those patterns turn into something usable). That’s a different kind of endeavour than most ML roles: less “optimize against a metric,” more “feel out a pattern nobody’s named yet, then prove it’s real.”

Voyage poetics.

• Strong deep learning fundamentals, with a feel for designing architectures around new inductive biases
• Sharp sense for representation learning and working with unlabeled data
• Ability to reason about signal-to-noise situations and integrate multiple modalities
• Experimental rigour and speed under high uncertainty – creating and testing hypotheses quickly, with direct oversight of hardware and data collection
• Experience building foundational models, designing tokenizers, and handling continual learning despite domain shift is highly useful

We care more about people who developed neurotech interest from an ML fundamentals background than people who started in neurotech and backed into modeling.

ML with us means self-supervised approaches to calibrating across modalities that were never designed to talk to each other. You’ll have a direct line to the hardware and data collection loop – when the data looks wrong, you’re close enough to the sensor stack to find out why, not waiting on someone else to tell you. The playbook for building in Mind Computing has not yet been written – you will be one of the first to do so.