Svyatoslav Korneev

I am leading the development of an AI workflow for multimodal medical imaging, focusing on detecting pulmonary embolism and building AI-assisted thrombectomy solutions. My responsibilities include model training and validation, establishing annotation infrastructure, and developing new algorithms for medical AI. I utilize the MONAI framework for AI development and Wolfram Mathematica for advanced image processing.

I conduct statistical and computer vision analyses of pathology slides to develop mathematical models that predict cancer treatment outcomes. This involves extracting key features from pathology images and constructing predictive survival models. I also calibrate models and design statistical tests to ensure their accuracy and generalizability.

I led the development of mathematical models and advanced image processing techniques to drive R&D in new additive manufacturing technologies. My role included collaborating with stakeholders, directing research initiatives, publishing academic papers, generating intellectual property, and mentoring junior research staff. I developed predictive machine learning models incorporating physical priors to analyze fluid flow experiments. In collaboration with Stanford University, I also created a framework for the symbolic homogenization of partial differential equations.

I engineered computational physics and image processing algorithms to analyze the oscillatory behavior of two-phase flows. Furthermore, I developed a reduced-order modeling framework employing convolutional neural networks to assess uncertainty in as-printed components. My expertise also includes multi-scale and multiphysics modeling, along with solving partial differential equations numerically using OpenFOAM and Wolfram Mathematica.

I developed an ML multi-scale framework for the prediction properties of Li-Ion batteries.

Development a numerical framework for approximating the solution of partial differential equations using binarized neural networks.

I develop advanced algorithms for image processing and segmentation of tomographic images from geological samples. I invented an efficient morphological filter concept to reconstruct unresolved pore spaces from X-ray computed tomography images of natural porous media columns. Additionally, I am actively engaged in creating novel multiscale computational methods for modeling reactive transport in heterogeneous porous media.

I investigated gaseous detonations in supersonic flows for various heat release models and published a paper in the most respectful journal, Journal of Fluid Mechanics (JFM (760), pp. 313-341, 2014, "http://dx.doi.org/10.1017/jfm.2014.598").

Research of nonlinear dynamic of Bose-Einstein condensate.

I developed a system for data collection and analysis for the Q&A department of a Gillette factory.