At Sberbank, I built the retrieval layer for an internal knowledge assistant used by support and operations teams. I implemented Python and FastAPI ingestion endpoints with Qdrant retrieval, and improved answer correctness from 76% to 84% on a reviewed evaluation set.
At Yandex, I managed delivery of ML-assisted text classification and content-routing features for internal operations. I owned roadmap, release scope, and acceptance criteria across four workstreams, and raised on-time milestone completion from 79% to 90% over four quarters.
I started as a Junior Software Developer at Yandex, maintaining Python API endpoints and data-processing jobs. I also coordinated delivery for Repetam, an online tutoring platform, and built projects involving LLM evaluation, document automation, and demand forecasting.

