Key responsibilities:- Consortium Participation: Contributing to the international consortium ‘CAR T Cells Rewired to Prevent EXhaustion in the Tumour Microenvironment’, aiming to enhance the efficacy and longevity of CAR T-cell therapies in solid tumors.- Efficiency Prediction Tools: Developing innovative tools to predict the efficiency of gene editing systems, ensuring precise and effective genetic modifications.- NGS Data Analysis: Creating and optimizing programs for the analysis of DNA and RNA Next-Generation Sequencing data, providing critical insights into genetic and transcriptomic landscapes.- HDR Donor Tools: Designing tools for the generation of Homology-Directed Repair donors, facilitating accurate genome editing and therapeutic interventions.- miRNA/siRNA Tools and Constructs: Designing tools for searching miRNA/siRNA, predicting knockdown efficiency, and making constructs of polycistrons and mirtrons.- Primer Searching Tools: Creating and refining tools for searching primers, essential for efficient and precise amplification of target DNA sequences.- Guide RNA Folding Optimization: Optimizing the folding of guide RNAs to ensure their stability and efficacy in gene editing applications.- Experimental Analysis and Assay Preparation: Analyzing the results of experiments and preparing assays to validate and support research findings.- Database Utilization: Utilizing databases to search and analyze large sets of genetic and experimental information.- Tool Development and Programming: Creating user-friendly tools using Flask and Docker to support scientists within the company, and developing tools using Python and R. Additionally, writing scripts in Bash and Nextflow to streamline and automate various bioinformatics workflows.My work is dedicated to improving the efficacy and safety of CAR T-based therapies, ultimately aiming to make a significant impact on patient lives by advancing the science of genetic engineering and immunotherapy.

Participation in the development search system for reference information based on transformer models:- I created pipelines for training the BERT NER model. It increased search quality by 230%. - I developed data augmentation algorithms that decreased the time for data labeling to 5 times and increased model metrics by 25%. - I optimized the model for the CPU. It decreased the time to get predictions by six times.

Participation in the development of software for reconstruction of 3D ultrastructure of presynaptic endings according to electron microscopy data:- Developed and trained models for detection and segmentation of the boundaries of internal compartments.- Developed algorithms and wrote program code for 3D reconstruction of detected objects and calculation of their characteristics.- 3D structure visualization software adaptation.My software allowed scientists to reduce data preparation time to 30 minutes (instead of 4-5 months) and increase the number of samples tested tenfold.

Developed and trained models and algorithms for the following projects:- Search system for reference medical and biological information based on transformer models. This reduced the search time by 300% and increased the number of relevant texts twice. - Chatbot for responding to complaints and suggestions. My pipeline increased the number of messages with answers tenfold and decreased waiting time from 24 hours to 10 minutes.- System for recognizing text information about a product from a photo in a store. My pipeline decreased monitoring time for one shop from 8 hours to 30 minutes. Also, it had 8% errors instead of 20% early.- Computer Vision system for recognizing safety violations by employees. My pipeline increased the number of identified dangerous situations by 75%.- Visual detection and localization of radiographic medical images. It helped to decrease the working time by 25% and increase detected diseases by 7%.- Predicting water levels for a network of hydropower plants. My model increased prediction metrics by 37%.- Software for calculating the effective selling price of an apartment. It increased the company's profit by 23%.Developed bioinformatics projects:- Predicting knockdown gene efficiency using shRNA. It decreased the count of failed experiments by two times and explored complex tasks that were unavailable early.- Automated pipeline for analysis of DNA data for patients with epilepsy. It decreased the time for data analysis by three times and found essential gene variants, which were unfamous early. - Automated pipelines and machine learning models for single-cell and bulk-cell RNA data from people with a heart valve calcification problem. It decreased the time for data analysis by five times and found essential genes.- Cardiac disease prediction according to ECG. It provided preclinical screening for Covid-19-related injuries and reduced the time to diagnose correctly by 15%.

Developed algorithms and machine learning models for forecasting product demand and optimal prices. My pipeline increased sales by 150-200% for some categories.

- Developed algorithms for the localization of defects in rolled products.- Post-processing of data and preparation for training the classifier.