Egor Kosaretsky
Bioinformatician / Data Scientist
Bio
Hi again! I'm a student at Lomonosov Moscow State University. I'm passionate about Machine learning and often hang out on kaggle searching for new datasets. I also like to apply my skills in computational and structural biology. In a free time i make simple indie games using python. Linux enthusiast and open source software contributor(I use Void btw).
Work at DevSect
Building AI infrastructure from ground up with help of Docker, AirFlow/Langchain, Kubernetes, PostgreSQL and FastAPI. Responsible for researching, finetuning Llama and other LLMs, text classificators to develop most efficient talking head pipeline enhancing user engaging in chatbot mobile app.
Work at GENXT
I have developed and held responsible for testing of pipeline of pedigree simulation with unrelated founders, relatedness degrees estimation and compare between true and estimated degrees. In my work I actively use ERSA, KING, IBIS, Ped-Sim with samtools inside Snakemake. Important part of my work is also pipeline speed enhancement, which requires computing power estimation and optimization. I'm also involved in Federated machine learning engine research for phenotype prediction based on SNV data. Working with XGBoost, Random Forest and MLP models fine tuning.
Work and Education
My skills
Languages:
Russian: Native, German: Upper-intermediate, English: Upper-intermediate, Japanese: Basic.
Programming skills and languages:
Python: Pytorch, NumPy, Pandas, Polars, scikit-learn, PyMol, SeaBorn, Luigi, Snakemake, FastAPI
Cloud and Databases: Google cloud, Yandex cloud, Docker, Kubernetes, PostgreSQL, Airflow, Luigi, Hadoop, Mlflow
R: tidyverse, ggplot2, caret
Web: HTML, CSS, JavaScript, FastAPI