Style Transfer Lab Team

Our Mission

Style Transfer Lab is a collaborative research group advancing the mathematical foundations of generative AI. We specialize in optimal transport, information geometry, and latent space analysis to develop both rigorous theoretical frameworks and practical applications.

Our research focuses on:

  • Optimal Transport Theory: Novel applications to color transfer, voice conversion, and image harmonization using Monge-Kantorovich solutions
  • Information Geometry: Fisher information metrics and geometric analysis of latent spaces in generative models
  • Latent Space Understanding: Revealing phase transitions, fractal structures, and geometric properties of diffusion models
  • Theoretical Foundations: Mathematical guarantees and convergence analysis bridging statistical physics and machine learning

We bridge theory and practice with applications including:

  • Biological Data Imputation: Reconstructing chromosomal distance matrices using generative methods
  • Lightweight On-Device Solutions: Real-time optimal transport algorithms for mobile and AR applications
  • Training-Free Methods: Zero-shot approaches that extend existing models without retraining

We publish at premier venues including NeurIPS, ICML, and AAAI, and are committed to open research with publicly available code and datasets.

Our Team

Alexander Lobashev

Researcher, Glam AI

Research interests include optimal transport theory, information geometry, and latent space analysis of generative models. Focuses on developing training-free methods and theoretical frameworks for diffusion models.

Research Interests: Optimal Transport, Information Geometry, Latent Space Analysis

🎓 Scholar 📖 OpenReview 💼 LinkedIn

Maria Larchenko

Researcher, Magicly AI

Research interests include color transfer, optimal transport methods, and lightweight on-device implementations. Specializes in efficient algorithms for real-time applications in AR and mobile platforms.

Research Interests: Color Transfer, Optimal Transport, Efficient Algorithms

🎓 Scholar 📖 OpenReview 💼 LinkedIn

Dmitry Guskov

Researcher, Glam AI

Research interests span generative models, latent space geometry, and applications of optimal transport to scientific and creative domains. Works on both theoretical foundations and practical implementations.

Research Interests: Generative Models, Latent Space Geometry, Optimal Transport

🎓 Scholar 📖 OpenReview 💼 LinkedIn

Why Work With Us

### 🎯 Expertise Across the Stack We combine theoretical depth with practical implementation skills, from mathematical foundations to production-ready code. ### 🏆 Track Record of Impact Our papers appear in top-tier venues (CVPR, ICML, AAAI) and address both fundamental research questions and real-world applications. ### 🤝 Collaborative Approach We thrive on collaboration, bringing together diverse perspectives to solve challenging problems in generative AI. ### 🚀 Research-to-Application Pipeline From initial concept to working implementation, we deliver complete solutions with reproducible code and clear documentation.

Collaboration & Opportunities

We are actively seeking opportunities for:

  • Research Collaborations: Joint projects with academic or industry partners
  • Consulting: Technical expertise for generative AI and computer vision projects
  • Full Lab Hiring: We’re available as a cohesive team for ambitious research programs

Areas of Expertise

🎨 Color & Style Transfer

Optimal transport, rectified flows, diffusion guidance

🤖 Generative Models

Diffusion models, GANs, training-free methods

📐 Theoretical Foundations

Latent space geometry, information theory

🔬 Applications

Fashion AI, creative tools, scientific computing

Last updated: February 2026