At BrokerDeck, I implemented a hybrid text-and-vector search engine for real estate properties. I also built a recommendation system using search infrastructure and property vector representations.
I built an LLM-based model to generate natural language comparisons between listings from structured property data, using Python, PyTorch, HuggingFace, and vLLM.
At Tinkoff Bank, I translated business needs into technical specifications for data storage. I developed SQL logic to clean, standardize, and join datasets from multiple sources.
With HackForLA, I create educational content for team members and community learners. I also worked as a teaching assistant for the Python programming basics course at Higher School of Economics.
