Hard problems require multiple forms of intelligence.
MNEOS is being built around eight founding research themes. Each is deliberately cross-disciplinary. Each requires the integration of computation, evidence, human judgment, and physical reality.
Eight areas of ongoing and prospective work.
Computational physics
Electromagnetics, waves, thermal systems, fluids, structures, multiphysics interactions, reduced-order models, and uncertainty quantification.
AI for physical systems
Scientific machine learning, surrogate models, physics-informed learning, evidence-aware agents, simulation orchestration, and model evaluation.
Robotics & embodied intelligence
Humanoids, mobile systems, manipulation, laboratory robotics, manufacturing assistance, autonomy, and human-machine teaming.
Materials & manufacturing
Computational materials, additive manufacturing, process modeling, ceramic systems, composites, qualification, and digital production.
Biological architecture mining
Sensing, adaptation, distributed control, robustness, pursuit, swarm coordination, immune-inspired detection, morphogenesis, and evolutionary search.
Advanced sensing & communications
RF sensing, distributed arrays, sensor fusion, resilient communications, contested-environment operation, and edge systems.
Intent-to-engineering
Voice-to-CAD, voice-to-RF, structured technical intent, interactive design, inverse design, and engineering-agent workflows.
Institutional intelligence
Memory, provenance, evidence chains, decision governance, permissions, technical knowledge graphs, and long-duration human–AI collaboration.
Six founding questions.
Provisional challenges that shape the initial research agenda. Each is deliberately larger than a single project.
Governed intent
Can engineering intent be translated into governed, testable design workflows?
Safe embodied AI
How can humanoid systems safely become useful scientific and manufacturing collaborators?
Evidence-updated models
How can physical test evidence continuously update simulation and AI models?
Reusable failure
How can engineering failures become reusable institutional knowledge?
Biological inspiration
How can biological architectures inspire resilient sensing and autonomous systems?
Compounding without leakage
How can portfolio-level technical learning compound without violating IP boundaries?