Anastasiia Sarmakeeva, PhD

Machine Learning Engineer · Computer Vision · Synthetic Data

📧 asarmakeeva@gmail.com
🔗 LinkedIn · GitHub

Summary

Machine Learning Engineer (PhD) with 6+ years of research and production experience in computer vision, medical imaging, and generative models. Published at MICCAI 2025 on large-scale synthetic medical imaging datasets. Strong background in geometry-aware ML, reproducible research, and production-grade PyTorch/C++ pipelines. Experienced in collaborating across research, software, and regulatory teams.

Technical Skills

Programming & Systems
Python, C++, Git, Bash, CUDA, Docker

ML & Data
PyTorch, JAX, TensorFlow, Hugging Face (Transformers & Datasets)

Scientific Python
NumPy, SciPy, Pandas, Matplotlib, Scikit-learn

Cloud & MLOps
AWS, CI/CD pipelines, model deployment, large-scale data processing

Domain Expertise
Computer Vision · 3D Medical Imaging · Generative Models (Diffusion, GANs)
Synthetic Data · Model Evaluation & Robustness · Reproducible ML Systems

Work Experience

U.S. Food and Drug Administration

Machine Learning Engineer
Silver Spring, MD · Jul 2024 – Aug 2025

  • Developed and deployed deep learning pipelines for 3D volumetric medical image analysis, including detection and segmentation for breast cancer diagnosis from CT and mammography data
  • Designed diffusion-based and inpainting generative models for medical image synthesis, improving segmentation model generalization via data-centric strategies
  • Co-developed and released T-SYNTH, a 10TB+ synthetic mammography dataset, with full documentation and public release on Hugging Face (presented at MICCAI 2025)
  • Built evaluation frameworks to assess synthetic vs. real data distributions, using statistical methods to analyze model robustness and failure modes
  • Designed scalable data curation pipelines for training generative models in production environments

George Washington University

Computational Research Developer
Washington, DC · Aug 2020 – May 2024

  • Led validation and verification of a CFD–DEM coupling solver for landslide simulations, designing experimental protocols and reproducible benchmarks
  • Contributed to an FDA–NSF research grant on credibility and reproducibility of computational modeling for medical devices
  • Awarded People’s Choice Award at the 3 Minute Thesis competition for effective communication of complex technical work

George Washington University

Teaching Assistant
Washington, DC · Aug 2021 – May 2024

  • Developed course materials and assessments for numerical methods and programming courses
  • Mentored students in Python, regression, data analysis, and numerical methods
  • Supported curriculum improvements and in-class instruction

Department of Particulate Flow Modeling

Computational Research Developer
Linz, Austria · Oct 2018 – Jun 2019

  • Developed approximation models combining Volume of Fluid and Discrete Element Methods
  • Presented computational approaches for Navier–Stokes–based fluid–structure interaction problems

Russian Academy of Sciences

Researcher
Izhevsk, Russia · Sep 2013 – Sep 2018

  • Conducted research in computational simulation of fluid flows and deformable bodies
  • Developed GPU-accelerated algorithms for sparse matrix inversion using CUDA
  • Worked on mesh-free and grid-based methods for fluid–structure interaction

Education

George Washington University

PhD, Mechanical and Aerospace Engineering
Washington, DC · 2019 – 2024

Udmurt State University

MSc, Applied Mathematics and Computer Science
2009 – 2014

Publications

  • Wiedeman C., Sarmakeeva A., et al.
    T-SYNTH: A Knowledge-Based Dataset of Synthetic Breast Images — MICCAI 2025

  • Sarmakeeva A., Barba L.
    Resolved CFD–DEM Coupling Method for Two-Phase Fluids Interacting with Arbitrary Shaped Bodies, 2024

  • Tartakovsky E., Plesovskikh K., Sarmakeeva A., Bibik A.
    Autocorrelation of Returns in Major Cryptocurrency Markets, arXiv, 2020

  • Sarmakeeva A., Tonkov L., Chernova A.
    Meshfree Methods for Fluid–Structure Interaction, 2017

  • Nedozhogin N., Sarmakeeva A., Kopysov S.
    Sherman–Morrison Algorithm for Sparse Matrix Inversion on GPU, 2014

Selected Talks & Presentations

  • FDA / MDIC Symposium on Computational Modeling and Simulation, 2024
  • International HPC Summer School, 2023
  • ParCFD, 2023
  • SciPy Conference, 2022
  • 3 Minute Thesis, People’s Choice Award, 2023

Honors & Programs

  • Fulbright Scholarship — Reproducible Research Training (Barba Lab, USA)
  • Women in Quantitative Finance Mentorship Program, Morgan Stanley
  • Ernst Mach Scholarship — CFD–DEM Research (Austria)

Other

  • SC23 — International Conference for High Performance Computing (Participant)
  • Patent Author: Sherman–Morrison Algorithm for Sparse Matrix Inversion on GPU