Преподаватель Python в IT-школу

Описание

Zarina Serikbulatova +77717633032 | zarina.serikbulatova@gmail.com | linkedin.com/in/zarina-serikbulatova SUMMARY Data Scientist with 5+ years across telecom, medtech, and e-commerce, delivering classical ML and deep learning solutions end-to-end, from problem framing and dataset design through production deployment and post-launch model support. Built and maintain production forecasting and customer-scoring models (demand forecasting for an e-commerce Private Brands division; churn prediction and customer scoring in telecom) with Airflow-orchestrated pipelines, iterating directly with business stakeholders after launch. Expanding into deep learning and NLP through hands-on work with embeddings, RAG systems, and LangChain agents. Comfortable across the analytics/ML stack: Python, SQL, Spark, ClickHouse, PyTorch. Open to relocation. EXPERIENCE Data Scientist, Autodoc | Feb 2025 - Present | Remote ● Owned end-to-end demand forecasting model for Private Brands (classical ML ensembles + time series), from problem scoping through production deployment ● Built an inbound forecast model for warehouses, supporting operational planning across the supply chain ● Partnered with demand planning stakeholders to monitor forecast performance and iteratively adjust models post-launch Data Scientist, Kazakhtelecom | Jun 2020 - Mar 2025 | Astana, Kazakhstan ● Built and maintained churn prediction models across B2C/B2B segments, full lifecycle from development through production monitoring and upkeep ● Assembled and cleaned datasets from raw, incomplete, multi-source data; surfaced actionable insights for business teams ● Coordinated with cross-functional teams to align model outputs with business operations and market dynamics ● Implemented a customer scoring model deployed into production Machine Learning Engineer, Spika Tech | Apr 2024 - Dec 2024 | Remote ● Built a signal filtering and preprocessing pipeline for ECG data as a production-quality Python tool ● Performed treatment, analysis, and error diagnostics on electrocardiographic signal data ● Developed an atrial fibrillation classification model achieving 90+ F1 score, including offline evaluation INDEPENDENT PROJECTS - NLP / LLM ● Built a RAG-based product recommendation agent for K-beauty products, combining retrieval over product data with LLM-driven recommendation logic (LangChain) ● Building multi-step, stateful agent workflows with LangGraph, with a focus on RAG pipelines and tool- using agents EDUCATION MSc in Statistics and Data Science, KU Leuven | Oct 2021 - Aug 2023 BSc in Computer Science, Nazarbayev University | Sep 2016 - Jun 2020 TECHNICAL SKILLS ● Programming: Python, SQL, R ● Machine Learning: classical ML and ensemble methods (XGBoost, LightGBM, CatBoost), time series forecasting, deep learning (PyTorch) ● NLP / LLMs: RAG systems, LangChain agents, embeddings, NLP pipelines (spaCy) ● Data and Infrastructure: Spark, Airflow (pipeline orchestration and productionization), ClickHouse, MySQL ● Practice: model interpretability and error analysis, experiment tracking and reproducible ML workflows, cross-functional stakeholder collaboration LANGUAGES ● English - C1/C2 ● German - B1