Frontend developer with 7 years of building large Vue.js single-page applications. I focus on frontend architecture, reusable components and migrations of big codebases.
Recent years: Vue 3, TypeScript, Pinia, TanStack Query. I run code reviews and mentor, negotiate API contracts with the backend team and write technical documentation for QA.
I build RepoBuddy on the side and lean on AI tooling for routine work. Happy to switch the stack to React or Angular.
Tooling and code generation
A TinyMCE plugin for LaTeX formulas — it let the company drop a paid solution
Template generators on hygen that cut the time to scaffold new code
A UI-kit library shared across projects, documented in Storybook
A custom ESLint plugin that keeps code standards in the team
Migrations and refactoring
A full Vue 2 → Vue 3 migration with zero downtime: Class Options API → Composition API
Moved state from Vuex to Pinia and introduced TanStack Vue Query for server state
Refactored legacy code and the architecture of internal services
Non-trivial integrations
Monaco Editor for editing code right in the browser
PNG rendering from Vue components both in the browser and on a Node.js server
Real-time features over WebSocket and Pusher.js
An npm library wrapping Yandex.Metrica with a type-safe API
Build and performance
Moved the build to SWC with code splitting and tree shaking — 40% faster builds
GitLab CI/CD with automated deploys and source maps shipped to Sentry
SPA render optimisation: load speed and interface responsiveness
Architecture
Feature-based architecture for a codebase of 2000+ files and 550+ components
Component-driven development with components built in isolation
Strictly typed API contracts in TypeScript
Team and collaboration
Regular code review and help with architectural decisions
API contracts agreed with backend at design time — fewer breaking changes, work goes in parallel
Your first pull request to any open-source repo. Faster.
Indexes a public GitHub repository into a typed knowledge graph and answers questions about it: what runs here, what matters, where a first contribution could safely land. Answers are grounded in the actual code and cite it.
AST extraction across TypeScript, JavaScript, Python and Go into a graph of entities and relations — Postgres + pgvector
A planner picks operators over the graph (find_symbol → get_callers → retrieve_code_chunks); the executor runs them deterministically
Answers stream with citations: every claim is linked to the entity or chunk it came from
Contributor's tour: entry points, hot files, safe first-PR zones, an auto-generated architecture diagram
Issue triage: checks whether the work is already done — by commits, live PRs and similar closed issues
MCP server: the same graph is addressable from Claude Code and Cursor