d
devdmytro

Dmytro Z.

@devdmytro

AI Agent Engineer specializing in Claude SDK, MCP, n8n, OpenAI

Denmark
English
About me
I ship production AI agents and integrations for SMB and SaaS clients. Real systems with auth, error handling, and clean code another engineer can maintain. Specialty: Claude SDK and OpenAI agent development. Stack: Claude API, OpenAI API, n8n, Telegram Bot API, FastAPI, Next.js, Python, PostgreSQL. I build: MCP servers for Claude Desktop, n8n workflows, AI voice receptionists, RAG with verified citations, Telegram bots, Chrome extensions. Open-source proof: github.com/extezzu/polypalace. Based in Denmark, EU timezone overlaps US business hours. Same-day responses.... Read more

Skills

d
devdmytro
Dmytro Z.
Offline • 
Average response time: 1 hour

See my services

AI Integrations
I will build a custom mcp server for claude desktop, cursor and ai agents
From$95 / projectMore details
AI Websites & Software
I will build a rag chatbot for law firms with verified citations on case law
From$150 / projectMore details

Portfolio

Work experience

Self-Help

AI Agent Engineer

Self-Help • Freelance

Feb 2021 - Present • 5 yrs 4 mos

AI Agent Engineer building production-grade integrations and automations for SMB and SaaS clients since 2021. Specialty: Claude SDK and OpenAI agent development. I ship real systems with auth, error handling, observability, and clean code another engineer can maintain six months later. Recent shipped work: PolyPalace MCP server (open source, MIT licensed) is a multi-source synthesis tool for Claude Desktop. Aggregates project files, browser bookmarks, and saved messages into Markdown summaries. Built with TypeScript and Python, ChromaDB-backed semantic search. Source code: github.com/extezzu/polypalace n8n workflow automations connecting WhatsApp, Telegram, Notion, CRM systems with Claude API as primary reasoning engine and OpenAI as fallback. AI voice receptionists for SMB verticals using VAPI, Twilio, Whisper transcription, and ElevenLabs for natural voice synthesis. RAG pipelines with verified citations for legal and medical document Q and A, using pgvector and Pinecone for hybrid retrieval. Telegram and Discord bots with python-telegram-bot and aiogram, integrated with Claude API for reasoning-heavy conversation flows. Chrome extensions with Manifest V3 compliance, Claude API integration, and Stripe billing for SaaS-ready deployments. FastAPI backends powering AI features in SaaS products, with proper auth, rate limiting, and observability. Stack: Claude API (Anthropic), OpenAI API (GPT-4, function calling, embeddings), n8n, Telegram Bot API, FastAPI, Next.js 14, TypeScript, Python, Node.js, PostgreSQL with pgvector, Redis, Docker. Hosting: Vercel, Railway, Cloudflare Workers, AWS. Approach: scope tightly, ship in milestones, write code another engineer can pick up six months later. Loom walkthrough before commit, working demo before final payment. Based in Denmark, EU timezone overlapping US East Coast business hours. Communication in English, Russian, and Ukrainian.