Hi, I'm

Dima Yefimov

>Full-Stack Developer
Dima Yefimov

Senior Full-Stack Developer with 9+ years of commercial experience (since 2017) and a deep technical background stretching back to university projects in 2013. I specialize in modern React/TypeScript/Node.js ecosystems, building performant, scalable web applications for major enterprise clients in retail, logistics, aviation, and food manufacturing industries. My career spans the full product lifecycle — from Figma design and backend API development to frontend architecture and CI/CD pipelines — with strong expertise in internationalization, PWA development, real-time systems, and team mentoring. Currently exploring AI-augmented development workflows with Claude Code, Cursor, and MCP integrations.

12+yrs coding
9+commercial
3companies
B2english
Master's Degree in Computer Science, 2018

My Journey

2025–2026

AI Agents, Claude Code & Current State

Starting in mid-2025, I became deeply interested in AI agents and modern models. I initially used Cursor extensively, then transitioned to Claude Code and Codex. I designed custom skills and sub-agents for various scenarios, and set up integrations with Figma MCP, Atlassian MCP, and other tools.

I also explored context management in LLM systems separately. Over time, we began developing our own AI solutions for designing flexible workflows. Recently, I've also been using OpenSpec as part of the SDD approach.

And here, at the beginning of 2026, this is my current technical journey — from first university projects and CMS to modern frontend/backend development with active immersion in the AI space.

Skills & Expertise

Frontend

HTML/CSS12y
JavaScript10y
TypeScript6y
Jquery6y
React5y
Webpack5y
SCSS5y
Vue.js4y

Backend

REST API6y
Node.js5y
PHP4y
Python3y
Laravel3y
MongoDB3y
NestJS2y
PostgreSQL2y

DevOps & Tools

Git10y
Docker5y
Webpack5y
Figma5y
AWS4y
CI/CD4y
Photoshop3y
Github actions2y

Testing

Vitest3y
React Testing Library3y
Jest3y
Supertest2y
Playwright0.5y

AI & LLM

Claude Code1y
Cursor1y
Codex1y
MCP-integrations1y
AI-agents and skills1y
OpenSpec0.5y
ExpertIntermediate

Projects

ArnieEnglish Telegram Bot

↗

A Telegram bot for learning English vocabulary through spaced repetition and intelligent feedback. Originally built with Python, then fully migrated to Node.js. Features a companion PWA that was initially developed in Vue 3 and later rewritten in React with GraphQL. The bot has been running continuously since 2020 and is still active.

  • ◆Implements Levenshtein distance algorithm for fuzzy matching of user input
  • ◆Uses Ebbinghaus forgetting curve for optimal spaced repetition scheduling
  • ◆Google Cloud Text-to-Speech integration for pronunciation practice
  • ◆Migrated from Python to Node.js (Telegraf) for better performance
  • ◆Companion PWA rewritten from Vue 3 to React + GraphQL
  • ◆Active and serving users since 2020
Node.jsTelegrafMongoDBReactGraphQLGoogle Cloud Text-to-SpeechPWA

CareerBoard

↗

A Jira-style job application tracker built to bring order to the job search process. Lets you manage applications as cards with custom statuses, monitor progress across the hiring pipeline, prepare and reuse cover letter templates, and get notified about stale applications. Heavily automated: paste a vacancy URL and the system auto-parses the company name, job platform, position title, and posting date — and suggests a matching cover letter template.

  • ◆Kanban-style board for tracking job applications with custom status columns
  • ◆Automatic vacancy parsing from URL — extracts company, platform, position, and date
  • ◆Cover letter template management with smart template suggestions per application
  • ◆Stale application detection to surface unresponded or forgotten applications
  • ◆Full pipeline visibility from first application to offer or rejection
Next.jsReact 19TypeScriptViteSupabaseTailwind CSS

Black Water War

↗

A real-time online Battleship game built as a full-stack monorepo. Players place their fleet on a 10×10 grid and take turns firing at the opponent's board — hitting a ship grants an extra turn, sinking all 10 enemy ships wins the game. Supports private room codes for friend battles and a spectator mode. Built with a NestJS backend on Railway and a Next.js frontend on Vercel, connected via Socket.io for real-time communication.

  • ◆Real-time multiplayer via Socket.io — hit detection, turn management, live game state sync
  • ◆Full-stack monorepo: Next.js frontend on Vercel, NestJS backend on Railway
  • ◆Private room codes for friend battles and spectator mode
  • ◆Authentication and user management via Clerk
  • ◆Statistics tracking: shots fired, hits, accuracy per game
  • ◆Demo mode available without account
Next.jsReactTypeScriptViteMantineSocket.ioClerkNode.jsNestJSTypeORMPostgreSQLVercelRailwaySentry

Solar System WebGL

An interactive 3D visualization of the solar system built with WebGL and three.js. A personal exploration project created during a six-month deep dive into WebGL technology, featuring a simplified but visually engaging model of planetary orbits and celestial bodies rendered on HTML5 canvas.

  • ◆Simplified 3D model of the solar system with planetary orbits
  • ◆Built with three.js library for WebGL abstraction
  • ◆Interactive camera controls for exploring the scene
  • ◆Part of a six-month study of WebGL and 3D web graphics
WebGLthree.jsJavaScriptHTML5 Canvas