Tatyana Pichugina,PhD
Machine Learning Engineer (Computer vision) at Invonto- Claim this Profile
Click to upgrade to our gold package
for the full feature experience.
-
Russian Native or bilingual proficiency
-
English Full professional proficiency
-
German Limited working proficiency
Topline Score
Bio
Credentials
-
Deep Learning Specialization
DeepLearning.AIJun, 2021- Nov, 2024
Experience
-
Invonto
-
United States
-
Software Development
-
1 - 100 Employee
-
Machine Learning Engineer (Computer vision)
-
Feb 2023 - Present
• Orchestrated end-to-end machine learning projects: established business-oriented performance metrics, initiated image dataset collection, performed data labelling, executed iterative model experimentation, deployed and monitored model. • Significantly enhanced model performance in semantic segmentation by optimizing deep neural networks (U-Net), achieving an exceptional increase in Intersection over Union (IOU) from 0.76 to 0.90. Employed PyTorch Lightning as a dynamic training framework to drive these advancements effectively. • Successfully deployed a trained model utilising Flask as a web application on an EC2 instance in AWS, ensuring efficient and user-friendly access to the model's capabilities. • Gathered, curated and refined image datasets, resulting in improved model generalization and accuracy. • Employed advanced computer vision techniques to accurately measure segmentation masks within images utilizing OpenCV • Collaborated closely with cross-functional teams and stakeholders to integrate machine learning solutions into practical applications Show less
-
-
-
Max Planck Institute for Evolutionary Biology
-
Germany
-
Research Services
-
1 - 100 Employee
-
Senior Researcher
-
Apr 2018 - Mar 2022
•Mined, analyzed and interpreted big over 100 Tb unstructured data (microscopy videorecording) for 3 decision-making research projects with multiple stakeholders. • Checked data quality and data collection problems using SQL, and Python API for Omero Server Data Base. • Designed Python (Jupyter) pipeline to check data quality, clean, and segment objects for thousands of microscopy images, to facilitate data analysis and reduce hundreds of hours of manual work. • Detected anomalies in instrumental data collection, identified the problem and developed an automated Python script to check the experimental setup. • Identified meaningful clusters using Scikit-learn Gaussian Mixture Model combined with PCA. • Tracked version changes and shared code for 3 projects using Git and GitLab. • Led Image processing discussion group for institute members, organized talks, invited guest speakers, and managed surveys. Show less
-
-
-
The University of Auckland
-
New Zealand
-
Higher Education
-
700 & Above Employee
-
Researcher
-
2012 - 2015
• Designed from scratch a 3D predictive model using the Monte Carlo simulation method for whole yeast genomes using C. • Performed over 3000 extensive calculations using Auckland University Supercomputer (NESI) using SLURM, and Linux. • Identified meaningful clusters of genetic elements in 3D genomes data set using R. • Conducted statistical analysis for over 1000 3D genome structures. • Designed 3D visualizations of genes, epigenetic marks, and replication origins positions using R. • Published 3 papers in peer-reviewed journals (NAR, Scientific Reports, Nucleus) in collaboration with interdisciplinary cross-country research teams. • Performed literature search and summarized findings for over 300 scientific papers. • Gave 3 presentations at international conferences. Show less
-
-
Education
-
2012 - 2015
The University of Auckland
Doctor of Philosophy - PhD, Molecular Medicine -
2008 - 2010
Novosibirsk State University (NSU)
Master of Science - MS, Chemical Physics -
2004 - 2008
Novosibirsk State University (NSU)
Bachelor of Science - BS, Physics -
2022 - 2022
DataCamp
Data Scientist Career Track
Community
