Віктор

VIKTOR SKOBELIEV
AI Automation Engineer | n8n | Python | LLM Integration
[відкрити контакти](див. вище в блоці «контактна інформація») | [відкрити контакти](див. вище в блоці «контактна інформація») | LinkedIn | GitHub

PROFESSIONAL SUMMARY
AI Automation Engineer with 6+ years in IT and 2+ years focused on AI automation, building workflow automations, LLM-
powered pipelines, and API integrations for business clients. Specializing in n8n, Make, Python, and modern AI APIs (OpenAI,
Claude, Gemini). Experienced in translating business requirements into reliable, scalable automation systems — from CRM
pipelines and Telegram bots to computer vision and multi-agent AI orchestration.

SKILLS
Automation & Integration AI & LLM Programming & Data
n8n (workflows, agents, RAG) OpenAI API (GPT-4, Whisper) Python 3 (automation, pipelines, CV)
Make / Integromat Claude API (Anthropic) SQL / PostgreSQL
Zapier (basic) Gemini API (Google) JavaScript (basic)
Webhook triggers & event flows Local LLM (llama.cpp, ONNX) Bash (basic)
REST API / OAuth 2.0 Prompt Engineering OpenCV / Pillow
JSON parsing & transformation Structured JSON output SAM / rembg / ONNX Runtime
Error handling & retry logic Chain-of-thought prompting Google Sheets (formulas, GAS)
API key & token auth Multi-agent orchestration Git / GitHub
Google Workspace API RAG pipelines Docker (basic)
Telegram Bot API Vector store (pgvector, Supabase)
Notion API LLM validation layers
Postman / API testing Context management

PROFESSIONAL EXPERIENCE
AI Automation Engineer | Confidential Client (NDA) January 2026 – Present
• Built a Figma-to-HTML email pipeline: developed a custom Figma plugin that extracts design data and assets using the
native Figma API (no paid token required), automatically uploads images to FTP, and assembles production-ready HTML
emails via n8n — eliminating manual email coding entirely
• Developed an AI-powered banner generator supporting arbitrary sizes using GPT Image-2 and Banana-2 models; users
describe content and select a theme, while the system assembles optimized prompts under the hood and renders pixel-
accurate banners at any specified dimensions
• Built a project estimation chatbot: ingests company knowledge base (services, capabilities) into a RAG pipeline, conducts
a real-time guided interview with the client using context-aware clarifying questions, and produces a structured verdict on
feasibility and hour estimates — replacing manual scoping calls
Data Specialist (SQL) | Telecommunications Industry (Contract) October 2024 – October 2025
• Cleaned, transformed, and prepared large telecom datasets using SQL and Python (Pandas, NumPy)
• Migrated data from PostgreSQL to cloud storage and built automated data-preparation workflows
• Automated ingestion and transformation pipelines to replace manual data handling
Freelance AI Automation Engineer | Self-employed January 2024 – February 2025
• Built n8n workflow automations integrating LLM APIs (OpenAI, Claude) reducing manual processing time by 60–70%
• Developed CRM payment verification pipeline: LLM extracts and validates payment data from emails, auto-confirms low-
risk transactions, escalates edge cases — reducing manual review by ~80%
• Implemented full Google Workspace integration (Calendar, Sheets, Gmail, Drive) via OAuth 2.0 for scheduling and
reporting workflows
• Built Telegram bot for client booking automation: LLM handles FAQ, checks Google Calendar availability, confirms
reservations automatically
• Developed competitor price monitoring: automated scraping → LLM parsing → Google Sheets report → Telegram alert on
price changes
• Delivered 8+ automation projects across e-commerce, services, and operations verticals
System Administrator | Rudana Television and Radio Company February 2019 – September 2024
• Maintained workstations, network infrastructure, and broadcast systems for 50+ users
• Managed user accounts, system performance monitoring, and hardware diagnostics
• Provided IT support and troubleshooting across internal departments

KEY PROJECTS
Project Normandy — Local Multi-Agent AI Orchestration
Built a desktop AI pipeline using Python, n8n, and local Llama 3.1 models to convert cross-stitch pattern images into
structured embroidery data. Stack: OpenCV, SAM (Segment Anything), rembg, ONNX Runtime (GPU), KMeans color
quantization in CIELAB space, Floyd-Steinberg dithering. Designed a two-agent validation layer: DataCritic agent analyzes
numerical pattern data, VisionAnalyst agent processes image metrics and produces final verdict with PDF summary. Custom
GUI built with CustomTkinter.
GitHub: github.com/Viktor-Skobeliev/Project_Normandy_XStitch

EDUCATION
Master's Degree — Computer Systems and Networks | 2018

LANGUAGES
Ukrainian: Native English: B1 — technical documentation, written communication