Artem Lobantsev

Experience

• 

  ITMO University

  Sept 2016 - Dec 2020

  Projects:• Dataset augmentation using GAN generated data (in team with https://www.linkedin.com/in/anton-klochkov-267459162/).My role was: research planning and leading, datasets research, and preparation. GAN training, approach validation. The publication is on preparation.Scope of the research: imbalanced datasets lead to worse training results in different ML models. We've conducted research on the potential benefits of data augmentation with generated data. The approach was tested on a lung cancer nodule imbalanced dataset with synthesized nodules.Stack: pytorch, pytorch-lightning, pydicom.• COVID classification on computed tomography scans.Research planning, raw data preparation, intercommunication with radiologists, deep learning model training, model production deploy.Stack: pytorch, aws, jupyter-notebooks, pydicom, FastAPI, Docker, etc.• Fusion of medical images and accompanying free-text reports.Research planning, data preparation, X-Ray images classification using CNN, text reports classification using BERT, transfer learning, multi-modal fusion. The publication is on preparation.Stack: PyTorch, BERT (transformers lib), CNN. Show less

  • Research Fellow

    Apr 2018 - Dec 2020

  • Postgraduate

    Sept 2016 - Dec 2020
• 

  BrainGarden

  Mar 2019 - Dec 2020

  Machine Learning Engineer

  Projects:1. Robotics, SLAM, self-driving: robot localization in outdoor environments, sensor fusion. Research and prototype models for the fusion of GPU, IMU, and Visual Odometry. Data preparation for a deep learning model, research neural network architectures, appropriate for sensor fusion, training a neural network to predict Kalman Filter parameters.Stack: pytorch, numpy, pandas, plotly / dash, LSTM, transformers, self-attention, Extended Kalman Filter, Lie groups, Invariant Extended Kalman Filter.2. Robotics, SLAM, self-driving: robot localization in indoor environments using stereo-cameras and visual odometry, sensor fusion.The fusion of vSLAM localization from multiple monocular-cameras set. Experiments with simulated and real data. Constant work with a foreign team.Stack: ROS, rospy, RTAB-Map, OpenCV, Extended Kalman Filter. Show less

• 

  Philips

  Dec 2020 - now

  Research and development of the AI algorithms for cardiology applications. AI in cardiology solutions

  • AI Research Scientist

    Sept 2022 - now

  • Data & AI Scientist

    Dec 2020 - Sept 2022