ML/AI and Data Engineer focused on content understanding, search/ranking, and large-scale systems.
I am a researcher and engineer working at the intersection of machine learning, aerospace systems, and computational geometry. My work spans large-scale ML systems for content understanding and recommendation, geometric modeling of complex structures, and data-driven modeling of physical and dynamical systems.
Currently, I lead R&D initiatives at Axya (Montréal), where I design and deploy production-scale ML pipelines, multimodal generative systems, and recommendation engines that power supplier discovery and process automation. In this role I work with major industrial customers in aerospace and manufacturing on data analysis, decision support, and automation.
I completed my PhD in Mechanical Engineering at McGill University with Prof. Yaoyao Fiona Zhao, focusing on geometric modeling and generative design for heterogeneous lattice structures and metamaterials. Before that, I obtained my MSc at the Skolkovo Institute of Science and Technology in a program developed under the MIT-Skoltech Initiative, with close ties to MIT AeroAstro and systems engineering.
I have also been a visiting researcher at the Airbus R&T Campus in Toulouse, ISAE-SUPAERO (Department of Complex Systems Engineering, with Prof. Rob Vingerhoets), and C3D Labs, working on model-based systems engineering, aerospace system modeling, and CAD kernel applications.
Earlier in my career, I collaborated with researchers at MIT and Skoltech on robotics and dynamical systems, working with computer vision, signal processing, and state estimation methods such as Kalman filtering for time-series and spatio-temporal data.
My broader interests include large-scale ML infrastructure, time-series and spatio-temporal modeling, data-centric AI, and generative design systems. For my parametric modeling software for cellular structures, see LatticeQuery.
McGill University, Montréal, QC, Canada
Thesis title: Geometric modeling of heterogeneous cellular structures with functionally graded parameters. Download.
Skolkovo Institute of Science and Technology, Moscow, Russia (MIT-Skoltech Initiative)
Thesis title: Integrating Concurrent Conceptual Systems Design with 3D Modeling. Download.
Bauman Moscow State Technical University, Moscow, Russia
Thesis title: Development of automated storage and retrieval system.
R&D Department
Visiting Researcher, Research and Technology Campus
Visiting Researcher, Department of Complex Systems Engineering (DISC)
Researcher, Space Center
Process Engineer
Mathematical modeling of multifunctional non-pneumatic tire designs. Read more.
PI: Prof. Yaoyao Fiona Zhao, McGill University
Mathematical modeling of composite metamaterials with high energy absorption and recoverability, and experimental tests with them. Read more.
PI: Prof. Yaoyao Fiona Zhao, McGill University ; Prof. Yi Wu, Tsinghua University
Investigated the friction factor and convective heat transfer coefficient across cellular structures. Read more.
PI: Prof. Yaoyao Fiona Zhao, McGill University
Development of methods and tools for the geometric modeling of heterogeneous cellular structures with function representation. Read more.
PI: Prof. Yaoyao Fiona Zhao, McGill University
Use of geometric and mathematical modeling to develop a model-based visualization approach. Read more.
PIs: Prof. Clément Fortin, Skoltech ; Prof. Rob Vingerhoets, ISAE-SUPAERO
A high-speed intelligent robotic system that improves the recycling process efficiency of electronic devices. The machines in this system can collaborate on dismantling electronics devices, retrieving and sorting valuable components from disposable ones. Read more.
PIs: Prof. Dzmitry Tsetserukou, Skoltech ; Prof. Kamal Youcef-Toumi, MIT
Power in electrical grids is modeled using a set of ordinary differential equations. The system is reduced to static one-dimensional flow, lacking branches, and containing an infinite number of loads. By converting a discrete model into homogeneous ODE's and rescaling ideal boundary conditions to initial conditions, distribution of active and reactive power, and voltage with respect to length are found. Read more.
PIs: Prof. Konstantin Turitsyn, MIT ; Prof. Ildar Gabitov, University of Arizona
Coding-wise, I mostly deal with Python for mathematical modeling, machine learning, data analytics, web-development, and engineering utilities; and with C++ for 3D modeling, computer graphics, CAD and other related software development fields.
You can explore my GitHub profile for more if interested.
nikita.letov [at] mail.mcgill.ca