Bio

I am a Generative AI Research Scientist working on diffusion language models as a faster, more controllable alternative to autoregressive LLMs, with broader interests in optimal transport and RL post-training.

I hold a Master's degree with honors in Math of Machine Learning from HSE University. I completed my bachelor's degree in Computer Science (Applied Mathematics) at NUST MISIS. I also have 3+ years of industry ML experience in RecSys, GNNs, and multimodal LLMs for robotics.

I am always open to discussing new ideas and enjoy sharing knowledge with a wider audience, so I founded the student ML community AI Knowledge Club, where you can also find my lectures.

Latest Updates

  • Oct 2026: New work: E-MoE! Accepted to the NeurIPS 2026 DiffuLM Workshop.
  • Sep 2026: TUBE was accepted to the NeurIPS 2026 (CORE A*) Main Track as a poster.
  • Jun 2026: Graduated from HSE University with a master's degree with honors!
  • May 2026: TCVBM and TUBE were accepted to the ICML 2026 SPIGM & FoGen Workshops as Spotlights. Excited to present them in person in Seoul, South Korea!
  • Feb 2025: Happy to attend the MenaML Winter School by Google DeepMind in Doha, Qatar!

Research Interests

  • Foundation of Generative Models (Diffusion, VAE, GAN, Flow Matching, Electrostatic models, etc.)
  • Sampling from Unnormalized Densities
  • Discrete and Continuous Diffusion Language Models, Multi-token Prediction, Speculative Decoding
  • Post-Training via RL-finetuning and Stochastic Optimal Control
  • Optimal Transport, Schrödinger Bridges
  • Looped Transformers, HRM, Latent Reasoning
  • Video Generation, World Models

Publications

E-MoE overview
E-MoE: Enhanced Mixture-of-Experts for Non-Factorized Diffusion Language Models
Arseny Ivanov | Alexander Kolesov | Alexander Korotin | Ivan Oseledets | Mikhail Goncharov
NeurIPS 2026 DiffuLM Workshop | arXiv, Code, Project Page, HF Papers
TUBE overview
TUBE: Tangent Upper Bound on Evidence for Discrete Diffusion Language Models
Arseny Ivanov | Sergei Kholkin | Vladislav Gromadskii | Grigoriy Ksenofontov | Ivan Oseledets | Alexander Korotin
NeurIPS 2026 Main Track (Poster) | arXiv, Code, Project Page
TCVBM overview
Time-Correlated Video Bridge Matching
Viacheslav Vasilev | Arseny Ivanov | Nikita Gushchin | Maria Kovaleva | Alexander Korotin
ICML 2026 SPIGM & FoGen Workshops Spotlight | arXiv
GCT-TTE architecture
GCT-TTE: Graph Convolutional Transformer for Travel Time Estimation
Vladimir Mashurov | Vaagn Chopurian | Vadim Porvatov | Arseny Ivanov | Natalia Semenova
Journal of Big Data, 2024 | arXiv