Andrei Panferov
AI PhD Research Intern at NVIDIA
Vienna, Austria
Summary
🤩
Rockstar🎓
Top SchoolAndrei Panferov is an ML-focused software engineer and PhD researcher based in Vienna with six years of experience building and optimizing large language models and ML infrastructure. He contributed core quantization work to Hugging Face Transformers—implementing AQLM support and CUDA graph compatibility—and co-authored research on extreme LLM compression while at Yandex. His background spans industrial and research roles at NVIDIA, Yandex, KAUST, and Terra Quantum, combining practical engineering (production YaLM work, quantum circuit tooling) with academic rigor. Currently researching LLM efficiency at ISTA, he brings deep expertise in model compression and efficiency that bridges code, experiments, and reproducible open-source teaching (contributions to the YSDA NLP course). An unconventional detail: he pairs high-impact library engineering with notebook-level improvements and benchmarks, reflecting both low-level systems skill and a focus on reproducible ML education.
7 years of coding experience
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Institute of Science and Technology AustriaMachine Learning Engineer, Machine Learning, Machine Learning Engineer, Machine Learning at Yandex School of Data AnalysisApplied Mathematics and Physics, Applied Mathematics and Physics at Moscow Institute of Physics and Technology (State University) (MIPT)High School Diploma, Physics, High School Diploma, Physics at AESC MSU — Kolmogorov Boarding SchoolDoctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at EPFL
Github Skills (18)
transformers10
pytorch10
python10
testing10
machine-learning10
deep-learning10
natural-language-processing10
cuda10
quantization10
nlp10
acc9
accelerator9
acceleration9
jupyter-notebook9
efficientnet7
Programming languages (6)
Github contributions (5)
huggingface/transformers
Mar 2023 - Oct 2025
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
Role in this project:
ML Engineer
ML EngineerContributions:52 reviews, 8 PRs, 91 comments in 2 years 7 months
Contributions summary:Andrei made significant contributions to the integration of the AQLM (Additive Quantization of Language Models) quantization method within the Hugging Face Transformers library. This involved implementing AQLM quantizer support, including AQLM configuration, and integrating it with existing model architectures. The user also developed and tested integration tests for AQLM, and made critical changes to enhance the existing CUDA graph generation for quantizers compatibility.
audioinferencemachine-learning-modelsmultimodaltransformers
yandexdataschool/nlp_course
Nov 2022 - Nov 2022
YSDA course in Natural Language Processing
Role in this project:ML Engineer
ML EngineerContributions:1 commit, 7 PRs in 1 day
Contributions summary:Andrei's commits focus on improvements and fixes within a natural language processing course. They addressed display issues within the `week06_da` homework, likely related to visualization. Further contributions included the correction of wget links in the `week10_quantization` and `week10_efficiency` notebooks, suggesting work related to dependency management or data loading within the course materials. They also contributed to the `compression hw` by adding `benchmark.ipynb` and editing several notebooks to improve homework instructions.
natural-language-processing
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.