All researchers
Researcher
Dr. Adrien Mercier
Adrien leads the architecture programme at Apex — the cross-modal fusion layer, the physics-aware MLP, and the discipline-head decoder routing. Before Apex, he led architecture work at a frontier AI lab and held a research role in geometric deep learning.
Recent
Recent papers and notes.
A slice of what Adrien has been writing. External papers link to the research log; internal notes link where we can.
- 01Cross-modal fusion in physics-aware backbones (Apex, 2026)
- 02Why MLPs in a transformer are physics-aware in our case (Apex internal, 2026)
- 03The case for sovereign foundation models (Holst & Mercier, 2026)
The open problem
“Can we share weights between CFD trajectories and RTL netlists without an adapter?”
Interests
- · Cross-modal weight sharing
- · Architecture ablations at scale
- · Discipline-head decoders
If you work on this — or have a strong view — write to research@apexworldlabs.com. We respond to every researcher inbound.
Work on a problem on this page.
Apex hires small numbers of unusually senior people. If you've worked on the problem the researcher above is open on — write to careers@apexworldlabs.com with a one-page brief.