Data Scientist
VEON (Beeline Kazakhstan)
VEON is a global digital operator. Beeline Kazakhstan provides mobile and fixed connectivity and digital services in Kazakhstan.
Feb 2026 — PresentBuilt and tested music recommendations, search ranking, B2B detection, and data pipelines, including a recommendation experiment with about 10.7% higher engagement.
• FOR_YOU recommendation system: Built and A/B tested a LightFM-based personalized music recommendation feed for the streaming product; the treatment group showed ~+10.7% engagement lift with the dislike rate roughly halved versus baseline.
• Search ranking: Designed an offline evaluation framework over head and tail query sets and improved the Elasticsearch ranking formula, eliminating severe tail-query failures with no regression on high-traffic queries.
AI/ML Engineer
Atlanti.ai
Nov 2024 — Feb 2026Improved multilingual speech synthesis and built reproducible data preparation pipelines, including a 30% rise in listener MOS ratings for expressive speech.
• Emotion & style preservation: Designed training with multi-speaker, emotion-tagged data. The model could reproduce 7 emotional styles (neutral, joy, sadness, anger, etc.), achieving a 30% improvement in listener MOS ratings on expressive speech compared to baseline.
• Data pipeline innovation: Built a custom preprocessing pipeline (alignment, phoneme normalization, emotion tagging, HF dataset publishing) that reduced dataset preparation time by ~40% and made experiments reproducible for future iterations.
Machine Learning Engineer
õzen: Stream
Feb 2023 — Dec 2024Built music recommendations and analytics pipelines, contributing to about 20% more listening sessions, 10% higher CTR, and lower project costs.
• Implemented cost-effective solutions, resulting in a 20% reduction in project expenses.
• Streamlined project workflows, enhancing overall efficiency by 25%.
• Led a team in successfully delivering a complex engineering project on time and within budget; developed and implemented machine learning models for user behavior prediction.
• Developed a Music Recommendation System: leveraged Python (Pandas, NumPy, scikit-learn, PyTorch, openl3) to build recommendation algorithms, increasing user listening sessions by ~20%.
Data Analyst
INN LAB
Jul 2021 — Feb 2023Analyzed data and coordinated cross-functional work on an educational platform, helping decision-makers improve features and user engagement.
• Analyzed large volumes of data to derive actionable insights.
• Coordinated the development of an educational platform, managing a cross-functional team and negotiating with stakeholders.
• Conducted data cleaning, preprocessing, and visualization in Python (NumPy, Pandas, Matplotlib, Seaborn).
• Delivered actionable insights to decision-makers, enhancing platform features and user engagement.
Junior Data Analyst
Optimacros
Optimacros develops enterprise planning software for budgeting, forecasting, and operational planning.
Feb 2020 — Jun 2021Analyzed ERP data and helped build inventory models, identifying stock anomalies worth over $120,000 and preventing write-offs.
• Cleaned and visualised multi-GB ERP data; built exploratory notebooks & Tableau dashboards for the finance team.
• Contributed to predictive inventory models alongside a 10-person analytics squad.
• Identified stock anomalies worth >$120k using seasonal ARIMA, preventing write-offs.
