gavrilenkods

Developed a Python backtesting engine for trading strategies, evaluating PnL, fees, trade counts and scenario-level metrics on historical data. Built and integrated an ML-based rug pull detector as a risk filter for token-trading strategies, reducing exposure to high-risk entries. Optimized engine architecture and result-analysis workflows, making hypothesis validation and strategy comparison faster.

Developed Go backend logic for TAC, an L2 layer connecting the TON/TVM ecosystem with EVM infrastructure through the TON App Chain bridge. Implemented sequencer synchronization, message processing and Web3 interactions for bridge and protocol components. Added TON highload wallet support, increasing operation processing speed and protocol throughput. Optimized the Web3 scanner, message executor and internal Go message-processing pipeline for stability and performance. Built Solidity proxy contracts and DeFi integrations for Curve, Midas, Yield, Algebra and other apps; used TypeScript/Hardhat for testing, deployment and upgrades. Contributed to protocol architecture and consensus-related mechanics: confirmations, message flow and L2 component interaction.

Led backend delivery for client web services, web games, Telegram Mini Apps and internal analytics tools. Built Python/Go REST APIs, business logic, SQL storage, external API integrations, Telegram and Web3 infrastructure. Created scrapers, automation scripts and data pipelines for collecting, cleaning and processing external data. Worked on ML/CV projects, analytics dashboards, data marts and dbt model optimization; decomposed tasks and coordinated delivery on selected projects.

Delivered a CV/ML system for pulp-flow analysis, raw-material quality assessment and impurity detection in industrial production. Trained and fine-tuned neural networks; prepared and annotated data, validated models and integrated experiment storage. Built camera and point-cloud analysis scripts plus automatic license plate recognition tied to declaration/accounting records.