Polina

About Me I am a results-oriented leader with expertise in data science, AI solutions, and software development, specializing in scalable models and data-driven strategies that enhance business efficiency. Passionate about mentoring teams and aligning technical initiatives with business goals, I actively contribute to the academic community and drive innovation across industries.

Get to know me!

Hey! I’m Polina Tanasevich, Innovative and results-driven professional with extensive experience more than 6 years in data science, analytics, and software development. I have a proven track record of leading multidisciplinary teams, designing scalable AI solutions, and aligning technical initiatives with business objectives. With expertise in advanced analytics, machine learning, and algorithm optimization, I deliver measurable impacts, such as revenue growth, cost savings, and operational efficiency.

I am passionate about mentoring and empowering teams, fostering collaboration across departments, and presenting complex insights to stakeholders to drive strategic decision-making. My work is complemented by academic contributions through publications and conference presentations, and I continually strive to integrate cutting-edge technologies to solve real-world challenges.

Ready to tackle challenging projects and drive innovation, I bring both technical depth and strategic vision to every endeavor. AI specialist with experience in managing business analytics projects and implementing SOTA algorithms.

I’m deeply passionate about state-of-the-art AI and regularly present at international conferences and publish in industry journals. Feel free to contact me here.

Contact

My Skills

Leadership

Team Management

Strategic Thinking

Business Alignment

Cross-functional Collaboration

Python

BERT

LangChain

HuggingFace

NLTK

Gensim

Natasha

PyTorch

TensorFlow

Neural Networks

SQL

Hadoop

Greenplum

PySpark

AirFlow

Gitlab

Git

Scientific Research

BI-systems

Projects Below are some of my projects, completed for personal purposes or as part of courses, showcasing my interest in applying machine learning to real-world challenges.

LLM Travel Agent

Developed a language model-powered agent for personalized travel planning, handling user queries, recommending destinations, and optimizing travel itineraries.

Case Study

Top-5 Article Summarization Agent

Created an LLM agent that retrieves and summarizes the top 5 articles on a given topic. Integrated with web search APIs, the agent generates concise summaries, enhancing information retrieval speed.

Case Study

Science I actively contribute to the advancement of AI and data science through research and publications in peer-reviewed journals and presentations at international conferences. My work focuses on AI-driven solutions, machine learning algorithms, and data analytics, with a particular emphasis on their real-world applications in business and industry. Additionally, I hold patents for innovative software solutions that enhance computational capabilities, demonstrating my commitment to pushing the boundaries of technology. ResearcherID: HZJ-6632-2023, ORCID: 0000-0002-4796-0215, Google Scholar: https://scholar.google.com/citations?user=6XguSygAAAAJ&hl=rus

Numerical analysis of the effective thermal properties and the stability for NTE metamaterials

This article explores metamaterials with a negative thermal expansion coefficient (NTE), where properties are defined by geometric microstructure rather than material composition. The effective thermal expansion coefficients are calculated through numerical solutions of thermoelastic boundary value problems using the finite element method, averaging strain over the material's periodic cell. Stability under thermal loads is also analyzed. Results show that adjusting cell geometry can achieve zero or highly negative thermal expansion, and the material remains stable across a wide temperature range.

Abstract

Patent 2020665248

Program of structural calculations "cae - fidesys" - calculation of effective elastic modulus of heterogeneous material, subjected to non-linear-elastic pre-loading, in three dimensional case (with consideration of non-linear effects)

Abstract

Patent 2020667788

Numerical evaluation program of effective elastic-plastic properties of heterogeneous material in two-dimensional case (SOFT)

Abstract

Solving the inverse problem of finding the geometry of a metamaterial cell using machine learning

Solving the inverse problem of estimating the geometry of an auxetic metamaterial cell based on its elastic properties using machine learning and the Fidesys strength package. Physical mesomechanics. materials with multilevel hierarchically organized structure and intelligent manufacturing technologies. Publisher: Institute of Strength Physics and Materials Science, Siberian Branch of the Russian Academy of Sciences.

Abstract

Megagrant

Competition for the provision of grants in the form of subsidies from the federal budget allocated for state support of scientific research conducted under the guidance of leading scientists in Russian educational institutions of higher education, scientific institutions and state scientific centers of the Russian Federation, 075-15-2019-1890, 12.2019 - 12.2021, performer

Abstract

Conferences I have presented at several international conferences, sharing insights on AI-driven solutions for geoanalytics, machine learning applications in business automation, and the integration of AI in financial institutions.

AI Journey 2024

Comming Soon ... https://vkvideo.ru/video-22522055_456244721

Abstract

Process Mining 2024

I'm used process mining (PM) and machine learning (ML) to analyze errors in employee-customer interactions within an application. By collecting and processing technical log data, she developed a binary classification model to identify errors, optimized the customer journey, and implemented targeted improvements. Key techniques included process discovery, performance analysis, and anomaly detection. This analysis reduced handling time by NDA minutes per error session and shortened the customer journey by five steps, resulting in an annual financial gain of over NDA dollars. The project established a framework for continuous improvement in user experience, enhancing conversion rates and reducing churn.

Abstract

Data Fest 2024

Multimodal GeoEmbeddings: Methods, Results and Implementation This study explores using multimodal geo-embeddings from event data to enhance forecasting in banking and fintech. By leveraging PyTorch-LifeStream and advanced techniques (CoLES, NSP, SOP), geographic locations are represented as vector embeddings, improving model accuracy by up to 16%. The approach involves spatial aggregation using H3 geo-hashing and embedding storage in Hadoop for scalable monthly updates, showing significant potential for context-rich, predictive analytics in spatial data applications. Link: youtube.

Abstract

XIII All-Russian Congress on Theoretical and Applied Mechanics

Solving the inverse problem of finding effective properties of an auxetic metamaterial using Machine Learning algorithms The study was supported by a grant from the Russian Science Foundation (R&D: grant No. 19-71-10008. Russian Science Foundation): Multiscale geomechanical modeling based on the spectral element method

Saint Petersburg, August 21–25, 2023 Organizers: Ministry of Science and Higher Education of the Russian Federation, Russian Academy of Sciences, Russian National Committee on Theoretical and Applied Mechanics

Abstract

Job experience I have led and contributed to high-impact projects, including developing AI-driven solutions for geoanalytics, automating log error monitoring, and optimizing LTV for revenue growth. My work spans scalable solutions across industries, driving business efficiency and operational improvements. (in progress ...)

Tech Lead of Data Science

Sber 2024 - Now
• Leadership and Team Management: Led a cross-functional team of data scientists, engineers, and analysts in developing and implementing AI-driven solutions to improve business efficiency across multiple domains. Provided mentorship to junior team members, improving onboarding efficiency and fostering career development.
• AI Solution Development: Spearheaded the creation of a geoanalytics AI model that increased the accuracy of location-based demand forecasting by 15% through advanced feature generation techniques and multimodal embeddings.
• Scalable Automation: Designed and deployed a model for automated error detection and classification in log data, reducing undetected errors by 7% and automating manual monitoring processes, resulting in a 30% cost reduction for verification. The solution was scaled to monitor systems across multiple departments.
• Strategic Contributions: Partnered with business leaders to align AI initiatives with organizational goals, streamlining decision-making processes and identifying high-impact areas for AI implementation.
• Publications and Presentations: Published research in peer-reviewed journals and presented findings at international conferences, showcasing thought leadership and advancing industry best practices.

Details

Senior Game Analyst

VK / My.Games 2021 - 2023
• Revenue Optimization: Developed predictive models for LTV (Lifetime Value) optimization, increasing monthly revenue by 12%. Collaborated with marketing teams to design and implement data-driven promotional strategies, boosting participant acquisition by 37%.
• Operational Efficiency: Automated analytics workflows using Python and SQL, reducing labor costs by 200+ hours per month for the analytics team.
• Scalability and Insights: Created a revenue forecasting system integrated with the company’s analytics platform, saving over 6% on content development costs and improving decision-making across departments.
• Collaboration and Communication: Regularly presented actionable insights to executives and cross-functional teams, enabling data-informed strategic decisions for marketing and product development.

Details

Core Developer C++

CAE Fidesys 2020 - 2021
• Algorithm Optimization: Enhanced computational algorithms for the Fidesys CAE suite, improving accuracy and speed by 40%.
• Software Efficiency: Designed new computing functions that reduced the execution time of the software core by 50%, enabling faster simulations for clients in engineering sectors.
• Strategic Contributions: Worked with leadership to identify key areas for software performance enhancement, leading to a 10% growth in the user base through improved product offerings.
• Collaboration: Partnered with the software development and testing teams to address critical bugs, ensuring timely resolution and deployment of new features.

Details

Research assistant

Lomonosov Moscow State University 2019 - 2021
• Innovative Solutions: Developed and patented two software solutions that expanded the capabilities of the Fidesys CAE suite, attracting a 10% increase in user adoption.
• Team Leadership: Assembled and led a team of seven for a startup competition focused on materials characterization analysis using the SaaS model. Delivered a functional prototype and business model, achieving finalist status.
• Cross-functional Collaboration: Engaged with suppliers and internal teams to resolve complex issues, ensuring seamless integration of new components and software updates.
• Strategic Vision: Identified emerging research opportunities, influencing the roadmap for new computational features within the CAE platform.

Details

Math tutor

Independent and Institutional Teaching (2018-2019, 2022):
• Individual Tutoring: Prepared high school graduates for the Unified State Exam in Mathematics, with students achieving high scores (87+) and securing admission to their desired universities.
• Institutional Teaching: Conducted advanced mathematics classes at the A.N. Kolmogorov boarding school of the Specialized Educational and Scientific Center of Moscow State University. Taught Mathematics to schoolchildren at the Faculty of Mechanics and Mathematics, Moscow State University, inspiring academic curiosity and excellence.
• Mentorship: Provided personalized support and mentorship to students, improving problem-solving skills and boosting their confidence in mathematical concepts.
• Curriculum Development: Designed and implemented structured lesson plans tailored to individual student needs, balancing theoretical rigor with practical applications.

Details

Hey, My name is Polina

Get to know me!

My Skills

Projects Below are some of my projects, completed for personal purposes or as part of courses, showcasing my interest in applying machine learning to real-world challenges.

LLM Travel Agent

Top-5 Article Summarization Agent

Numerical analysis of the effective thermal properties and the stability for NTE metamaterials

Patent 2020665248

Patent 2020667788

Solving the inverse problem of finding the geometry of a metamaterial cell using machine learning

Megagrant

Conferences I have presented at several international conferences, sharing insights on AI-driven solutions for geoanalytics, machine learning applications in business automation, and the integration of AI in financial institutions.

AI Journey 2024

Process Mining 2024

Data Fest 2024

XIII All-Russian Congress on Theoretical and Applied Mechanics

Tech Lead of Data Science

Senior Game Analyst

Core Developer C++

Research assistant

Math tutor

Contact I’m always open to connecting with professionals and enthusiasts in AI and data science! Whether you’re interested in discussing AI solutions, collaboration opportunities, or just want to say hello, feel free to reach out through LinkedIn, GitHub, or via email.