I'm a Lead Software Engineer at Yandex. I work on YDB's web interfaces and developer tooling, mostly in TypeScript, and I'm one of the code owners of ydb-embedded-ui, the web UI built into YDB.
In open source, I build tools for vector search, local database operations, CI, and coding agents. I spend a lot of time on the boring parts too: tests, reproducible setups, failure modes, and permission boundaries.
- ydb-qdrant: a Qdrant-compatible vector search service and TypeScript library backed by YDB.
- local-ydb-toolkit: a Codex skill and MCP server for operating Docker-based local YDB deployments.
- my_instructions: project-neutral instructions for coding agents, with an eval harness that tests whether they actually change agent behavior.
- gravity-ai-ui: a UI generator that validates model output and renders it using trusted Gravity UI components.
- setup-local-ydb: a GitHub Action for provisioning disposable local YDB tenants in CI.
- codex-pets: a YDB-backed gallery and MCP service for Codex-compatible animated pets.
Most of my open-source work is in YDB Embedded UI: 300+ merged PRs covering diagnostics pages, the query editor, the AI assistant integration, component APIs, and Playwright e2e infrastructure. I also send occasional C++ fixes to YDB core (KQP query statistics, HTTP/TLS startup).
Elsewhere: Gravity UI, codebase-memory-mcp, and the codebase-memory-mcp skill in GitHub's awesome-copilot, which I added and maintain.
I wrote how semantic search works in Codex Pets. It covers embeddings, hybrid ranking, YDB storage, and fallback paths.