Adrian Kay is an engineer turned data scientist who has spent six years inside Madison International Realty, a global real-estate private-equity investor — building the AI models, machine-learning deal-ranking, and data infrastructure an institutional shop actually runs on. This isn’t theory; it’s the day job.
Before that he sourced and underwrote NYC acquisitions at Megalith Capital and built their ground-up algorithmic deal-sourcing platform. He started as a civil engineer at Arup (MEng, UCL), killing manual design work with Python and VBA — the same instinct, a different domain.
What’s rare is the range. From the GPUs, NAS, and networking in his own self-hosted AI lab down to the models, pipelines, and agents running on top, he works the full stack — hardware to software. Most teams have one side or the other; his systems reach production because he owns both.
Jarvis Labs is how he brings that in-house capability to operators who can’t staff a data-science team: a phased AI operating platform — deal flow, underwriting, inbox, then the IT backbone — built from the ground up around how you already work.