I began in mechanical engineering, grounded in physical systems and math. Today I connect those foundations with AI to build modular, constraint-aware systems that solve real problems.
I start with first principles: physical laws and math. Frame the system, set constraints, and design for control and clear prediction.
I carry those principles into AI with hybrid, constraint aware models. Reduce cost, shorten experiments, and keep learning stable and explainable.
Then I ship. Start with clean packages and repos, run through CI/CD with versioned artifacts and automated checks, and deliver observable services. Tight latency and cost, modular and scalable.
Modularity is a habit. Small components with clear interfaces make change safe and progress steady.
Engineering taught me to think in systems. In AI, I design components that align, exchange the right signals, and serve a larger goal. When they stay in tune, the system earns trust under real load.
Modular AI Systems
Physics-ML Integration
Production Scaling
Team Collaboration
Small parts. Clear interfaces. Compounding impact.
From physics foundations to modular AI systems—building the future one component at a time.
Grounding in physical laws, applied math, and computational modeling.
First-principles problem framing and model validation
Numerical methods: discretization, convergence, and error control
Multi-physics modeling using advanced numerical techniques.
Coupled magneto-hydro-dynamics modeling for ferrofluid micro-mixing
Mentorship and clear technical communication
Physics-guided ML for complex fluids with an emphasis on reliability and sample efficiency.
Proposed rheology-informed ML models for soft matter
Open-sourced tooling for training and evaluation
Ph.D. merit award; invited talks and conference presentations
Building AI systems for industrial R&D from lab to production.
Building state-of-the-art models for physical discovery
Agentic, industrial pipelines that scale and address meaningful bottlenecks
Modular solutions that bridge physics, AI, and production systems.
Industry-scale agentic systems leveraging cutting-edge technology for autonomous decision-making in complex physical environments.
Impact: Industry-scale value generation
Physics-informed ML models for design, discovery, and identification of soft complex matter.
Impact: Up to 91% reduction in experimental overhead
Learning families of constitutive models using neural operators.
Impact: Robust digital twins to costly rheological experiments
Multi-fidelity models for characterizing J&J baby shampoo with limited data.
Impact: 12+ months saved in R&D cycles
Explore my research publications and citations
Google ScholarA multidisciplinary skill set spanning physics, AI, and production systems.
Ready to collaborate on the next breakthrough in physics-informed AI? Let's discuss how we can make systems better together.
© 2025 Milad Saadat. Crafted with precision and passion.
Making physical systems better, one algorithm at a time.