Vlad Savinov
Staff DL Engineer & Team Lead, Pretraining @ Yandex
I lead the YandexGPT pretraining team at Yandex, focusing on distributed LLM training, optimizations and model architecture. Key work includes FP8 training recipes, Context Parallel for long-context models, MoE training at scale, and company-wide distributed training framework development. Previously led Applied ML team and worked on large-scale fine-tuning for enterprise search and code assistance.
Interests
- Distributed Training & Efficiency
- Reinforcement Learning
- AI Safety
Public Talks & Appearances
2025
- Inside LLM Pre-training: Scaling AI to Billions of Parameters - Podcast (December)
- LLM Scaling Week - Session 1, Session 2 (November)
- Speeding up Training with FP8 and Triton - YerevaNN + Yandex Hall, Yerevan (English) (November)
- The Technology Behind Large Language Models and GPT - Armenian Science Week, Yerevan (English) (October)
- Moscow State University - Efficient DL Systems (Russian) (September)
2023
- DataFest 2023 - Machine Learning Talk
- EnterMedia Interview - “Higher IT League”
Education
BSc in Mathematics, Algorithms and Data Science St Petersburg State University, 2019 - 2023
Thesis: “Learnable decentralized MAPF using reinforcement learning with local communication” (PDF) Supervised by Konstantin Yakovlev
High School Physics and Mathematics Lyceum No. 239, 2015 - 2019
Background
Before Yandex, I worked with the International Laboratory of Game Theory and Decision Making @ HSE on college admissions research, conducted RL research for traffic light optimization, and interned at Tinkoff as an Analyst.
I graduated from Physics and Mathematics Lyceum No. 239 in 2019, one of Russia’s leading schools.
Connect
- GitHub: acforvs
- LinkedIn: vladislav-savinov