At Verme, I designed the target architecture and development plan for a corporate LLM platform adopted by 15 users across three product teams. It enabled up to 5× faster completion of common tasks than manual workflows.
I also built MCP servers and agent skills for internal systems and data, with Human-in-the-Loop controls for potentially unsafe actions. In my earlier Python Backend Developer role at Verme, I refactored payable-time calculation and workforce-scheduling modules for major clients.
At ITMO University’s Research Center “Strong AI in Industry,” I developed and reviewed code for GOLEM, FEDOT, and GAMLET. I also contributed to a peer-reviewed AutoML paper and developed API endpoints for an urban-analysis LLM assistant using RAG.
Earlier, at Way2Win, I enhanced and supported a training-event platform with Telegram and Zoom integrations. At Portal-Yug LLC, I developed and implemented a lease accounting subsystem that saved approximately 100 person-hours per month.
