I build AI products end-to-end — from LLM pipelines, RAG and agents to the backend, data layer, and UI that ship them. One engineer, full cycle.
I'm a full-stack engineer who specializes in turning AI capabilities into real, working products — systems people actually use, not prototypes that demo well and break in production.
My daily work blends LLM integration (OpenAI, Gemini, Claude), RAG pipelines, embeddings and automation with a solid full-stack foundation: TypeScript, Next.js, Node.js, PostgreSQL. I've led a CRM team, built realtime systems with WebSockets, and turned plenty of unreliable third-party data into something dependable.
What I'm good at is the full cycle — architecture, the AI layer, the backend, the database, the interface, the deploy. The breadth that lets one engineer own a product from idea to result.
Custom analytics and payroll platform built on top of Alteg.io CRM with proprietary business logic not covered by the standard product.
AI-powered SaaS for automated church content generation.
Hybrid no-code/full-code wishlist platform and gift marketplace with automated reminders and recommendations.
Specialized SaaS platform for deceased estate administration in Australia.
Internal wiki platform with three-level data hierarchy.
Responsive web application with unified design system.
Own AI-powered SaaS product in the marketing and brand strategy space.
Full-featured booking platform for a private Ukrainian carrier operating Mercedes Sprinter routes between Ukraine and Western Europe.
Lightweight n8n / Make workflows built alongside product work — content and data pipelines, not standalone products.
An n8n workflow that scrapes a landing page URL and returns a structured AI analysis.
An n8n workflow that turns AI-generated task descriptions into structured tasks in Notion.
An n8n RAG pipeline that uses Qdrant and Gemini to answer user questions with context.
An n8n workflow that auto-generates content and posts it to a Telegram channel.
Centralized n8n orchestration of OpenAI chat, files, transcription, and audio generation.
An n8n workflow that turns uploaded PDFs into searchable embeddings in Qdrant.
A Make scenario that backs up Telegram files to structured Google Drive folders.
A Make webhook bridge that sends external files into Bubble as attachments.
Own product from idea to release: architecture, AI layer, and what to build next.
Building AI-powered SaaS, RAG assistants and LLM automation hubs alongside production full-stack work.
Led the team building a realtime CRM; owned auth, access control and sprint planning.
Shipped transport, medical and marketing platforms — Next.js, TypeScript, Node.js, Supabase.
Always open to interesting problems, collaborations, and conversations about building AI products end-to-end.