ResearchWorks LTD · Part-time Project · Tashkent, Uzbekistan

Python, LightGBM, SQLite, Statistical Analysis

• Developed behavioral risk prediction system for proprietary day traders using machine learning to predict negative PnL with 70%+ precision and Sharpe ratio over than 0.5.
• Engineered 20+ behavioral features from trading data (revenge trading, loss aversion, position sizing) with proper temporal validation to prevent data leakage.
• Built end-to-end pipeline: PropreReports API ingestion → SQLite database → feature engineering → LightGBM model → automated email alerts with calibrated confidence scores.
• Implemented time-series cross-validation and backtesting framework with transaction cost modeling, achieving lower Expected Calibration Error for risk predictions.

Optifly - Airline Scheduling Software · Dublin, Ireland (Remote)

Java, Python

• Applied advanced heuristic and metaheuristic techniques (e.g. Genetic Algorithms, TreeSearch Heuristics) to streamline airline scheduling workflows, reducing performance bottlenecks and elevating profit and speed performance at least by 10% and 25% respectively.
• Engineered high-priority constraint programming features within demanding client deadlines, fortifying both system reliability and operational throughput in combination of Java and Python languages.
• Led in-depth performance analyses to pinpoint and resolve bottlenecks, software defects, and suboptimal solution paths, driving improvements in runtime efficiency and overall solution quality.
• Rapidly learned airline industry domain intricacies, translating client expectations into effective, data-driven functionalities that exceeded operational targets.
• Explored and introduced advanced mathematical techniques (e.g., refined constraints, multi-objective optimization) to significantly boost performance and accuracy, while preserving clean, maintainable code.

Innopolis University · Innopolis, Russia

• Organized programming competitions, delivered lectures on advanced algorithms and trained university teams for the ICPC (International Collegiate Programming Contest), leading to qualifying of a team to World Finals

Huawei R&D, AI Lab · Moscow, Russia (Remote)

Python, C/C++

• Contributed to the development of a reinforcement learning framework (inspired by MuZero) for strategic decision-making, identifying optimization opportunities through comprehensive code-base analysis and profiling.
• Enhanced computational efficiency by migrating to high-performance data structures, accelerating core operations and throughput i.e. replacing segment tree data structure used in framework with faster alternative.

Upwork · Freelance

Python, C/C++, ML/DS

• Analyzed client data with Pandas and NumPy to spot bottlenecks, then tuned model parameters for more accurate, efficient results.
• Designed and built custom optimization algorithms and optimization in Python with OR-Tools/Gurobi, improving solution quality of the different projects
• Maintained a 100% job-success score and 5-star Top Rated Plus ratings through clear communication, on-time delivery, and proactive updates