Biography
I’m a research scientist at Yandex Research. My research interests include natural language processing, efficient deep learning and federated learning. I’m currently a final year bachelor’s student at Moscow Institute of Physics and Technology (MIPT).
Interests
- Natural Language Processing
- Efficient Deep Learning
- Federated Learning
Education
PhD in Computer Science, 2024-now
ISTA
BSc in Applied Mathematics and Physics, 2020-2024
Moscow Institute of Physics and Technology
Machine Learning Engineer, 2021-2023
Yandex School of Data Analysis
Experience
Senior ML Engineer (NLP)
- Overseeing the LLM deployment
ML Research Scientist
- Wrote a first-author paper on LLM Compression
- Achieved state-of-the-art results on LLM compression, reducing model size by 87% with acceptable loss in performance
- Wrote efficient inference kernels using Triton and C++, speeding up LLM inference by up to 320%
- Integrated the framework into the transformers library, enabling low RAM dispatch and reducing instance RAM requirements by 70%
Research Intern
KAUST, Optimization and Machine Learning Lab
- Conducted research under the supervision of Prof. Peter Richtárik.
- Authored a first-author paper on Correlated Quantization
Infrastructure Developer
Eqvilent (HFT Fund)
ML Engineer Intern (NLP)
Yandex
- Enabled abstract tabular data insertion for efficient map-reduce LLM inference, speeding up the tabular data processing by 120%
- Increased test coverage of the map-reduce inference interface from 0 to 85% through rigorous unit testing
Researcher
- Researched quantum algorithms for business applications.
- Developed an NMR spectra analysis toll, allowing for its use for for quantum computations.
- Optimized LLM deployment for chat assistant applications, reducing latency by 40%.
Accomplishments
International Physics Olympiad
Gold Medal
Projects
*