Work

DucensAI - CTO

George Babakhanov
George Babakhanov
Aug 31, 2025 · 1 min read

As CTO of DucensAI, I led the company's evolution from a marketing agency (dcnsmedia.com) to an AI automation services company.

Phase 1: Marketing Agency

January 2024 – Early 2025

Helped over 20 automotive companies grow their newsletters and client base. Free marketing case studies available at dcnsmedia.com.

Phase 2: AI Automation Services

Mid 2025

Transitioned to building AI-powered automation systems for businesses.

Key Projects

AI Invoicing Automation Designed an AI-powered invoicing system with backend PDF generation and CRM data sync, saving clients 10+ hours per week.

Legal & Medical AI Assistant Built a private AI assistant using local RAG over case law and medical databases, reducing search latency from hours to seconds.

AI Marketing Analytics Engine Developed an analytics engine leveraging transformer architectures for content and performance insights.

Fine-tuning & Infrastructure Integrated LoRA/QLoRA fine-tuning, AIBrix distributed inference, and SQL-linked pipelines for real-time reasoning and automation. If you want to see some sample code from the projects, there's a small part of it on GitHub.

Timeline
AI Invoicing Automation
Aug 1, 2025 - Aug 31, 2025
Designed system with PDF generation and CRM sync, saving 10+ hours/week
LoRA/QLoRA Fine-tuning
Jul 15, 2025 - Aug 31, 2025
Integrated AIBrix distributed inference and SQL-linked pipelines
AI Marketing Analytics Engine
Jul 1, 2025 - Jul 31, 2025
Developed transformer-based content and performance insights system
Legal/Medical AI Assistant
Jun 1, 2025 - Aug 15, 2025
Built private AI using local RAG over case law and medical databases
Marketing Agency Phase
Jan 1, 2024 - Mar 31, 2025
Helped 20+ automotive companies grow their newsletters and clientele
George Babakhanov
Written by
George Babakhanov
George Babakhanov is an engineer working at the intersection of artificial intelligence, systems, and real-world infrastructure. He builds reliable AI-driven systems, from model training and automation pipelines to fault-tolerant software and hardware integration. His work focuses on making complex systems understandable, deployable, and useful.