Pranay Seshadri

About me.

I live in London and work at the University of Cambridge and The Alan Turing Institute. I studied computational engineering at Cambridge and aerospace engineering previously at Maryland. During my doctorate I worked at Rolls-Royce plc in Derby and spent time at Stanford. My research focuses on uncertainty quantification, optimisation and physics informed machine learning to address challenges in aerothermodynamics. I am the founder of Effective Quadratures.

My current research projects (and interests) are:

• Bayesian models for improved aerothermal flow-physics in engines.
• Subspace-based dimension reduction: algorithms and 3D blade design.
• Polynomial approximations in uncertainty quantification, machine learning and optimization.
• Empowering next-generation digital twins in virtual reality.
• Machine learning techniques for extracting data-driven physical insight in flow simulations.

For all my publications, please check out Google Scholar and Researchgate. For open-source codes, visit my Github profile. For my CV, click here. If you are really interested in my research, check out the videos below. The first one is a high-level research talk, while the other two are initial results of my team's ongoing virtual reality work.

Team.

Ashley Scillitoe (Postdoctoral fellow): Physics informed machine learning in computational fluid dynamics.
Chun Yui Wong (PhD student): Embedded ridge approximations for improved computational aerodynamic inference.
James Gross (PhD student): Techniques in polynomial subspace projections for design optimisation.
Slawomir Konrad Tadeja (PhD student, co-advised by Per Ola Kristensson): Digital twinning an aeroengine in virtual reality.

Latest preprints.

• Embedded ridge approximations
• Gaussian ridge functions
• Higher order sensitivities via polynomial least squares
• Frequentist strategies for aero-engine spatial field approximation Part I
• Frequentist strategies for aero-engine spatial field approximation Part II
• Multipoint fan design with dimension reduction
• AeroVR: An immersive visualization system for aerospace design and digital twinning in virtual reality

Upcoming conference presentations.

• AIAA SciTech 2020: Optimisation with intrinsic dimension reduction: A ridge informed trust-region method.
• IEEE Aerospace 2020: Digital twin assessments in virtual reality: An exploration study with aeroengines.

Upcoming invited talks.

• NAFEMS European Conference: Simulation-Based Optimisation

Former team members.

Joe Zhou (Undergraduate student): High-dimensional quadrature rules.
Shaowu Yuchi (Undergraduate student): Data-driven dimension reduction.

Contact.

IN2-17 Inglis Building,
Department of Engineering,
University of Cambridge.
p.seshadri @ eng.cam.ac.uk | pseshadri @ turing.ac.uk