Ivan Anokhin
PhD student at Mila / Université de Montréal.
I am a PhD student at Mila / Université de Montréal, advised by Irina Rish. I am driven by a fundamental interest in the nature of intelligence. My broader research spans scalable reinforcement learning, memory mechanisms, continual learning, and (multi)-agency.
My current core PhD research applies asynchronous computation to design new architectures and pipelines that break the sequential layer execution of standard neural networks. The goal is to build scalable, wall-clock efficient, low-latency agents that are closer to biological intelligence.
Before Mila, I was lucky to work with Victor Lempitsky at Samsung AI Center and Dmitry Yarotsky at Skoltech. I also worked at Yandex and earned my BSc from Saint Petersburg State University.
Links: Email · CV · Google Scholar · GitHub · Twitter/X · LinkedIn
News
- 2026-03 - Will join Sakana AI as an intern for summer 2026.
- 2026-02 - Supported by the Université de Montréal AI scholarship.
- 2026-01 - Learning From the Past with Cascading Eligibility Traces was accepted to ICLR 2026.
- 2026-01 - Published my first blog post: Why Civilization Goes Virtual.
- 2025-12 - Passed Predoc-3 (mid-PhD committee exam). Topic: “Toward Asynchronous Computation for Streaming RL.”
- 2025-10 - Launched this website! 🎉
- 2025-07 - Preprint on bio-plausible learning rule that handles delayed credit assignment is out! 🎉
- 2025-07 - Mila blog post on real-time RL is out.