• Data engineering tasks (ETL pipelines)• Implementing services: trading profit/loss calculators, execution collectors, real-time data feeds, etc. • Implementing components of trading framework • Implementing trading strategies (translating research code to production code) • Implementing machine learning pipelines

• Backend (elixir, postgresql) and frontend (angular) development of CRM and customer web interface. • Developed mobile application from scratch in dart/flutter.

• Postflop solver (C++, cuda): calculates strategies and values in the game tree starting from flop.• Framework for poker development (collection of libraries in C++ with python ports): ◦ Machine and human representation of the game and hands. ◦ Hand comparison and equity calculation ◦ Generic builder of the game tree ◦ Generic solver of the game tree (requires EV evaluation on the leaf nodes) ◦ EV calculator for the river leaf nodes (CPU and GPU versions) ◦ Representation of players’ tendencies• Complete game model (python, tensorflow). Takes game state, returns suggested move. Takes into account opponent’s tendencies both preflop and postflop. Solutions in subsecond time. Preflop – empirical rules. Postflop – neural network model. Runs as a web service (play server).• Poker bot (in rust). Reads game events from the screen, maintains game states, interacts with the play server to get solutions with additional information (ranges and tendencies), saves them to db and broadcasts via websocket.• Web gui (next.js/react, redux). Displays the game in real time or historically with players’ ranges and tendencies as well as suggested moves for the hero with explanatory information.

Large-scale captcha solver: patched browsers, custom browser control library, control of browser fingerprint, full ML-pipeline.