Building Your First CrewAI Multi-Agent System
A comprehensive guide to creating intelligent agent crews that collaborate to solve complex problems. Learn the fundamentals and best practices.
Read More →AI Engineer & Autonomous Agent Developer
I architect autonomous agent systems that streamline your business operations. Leveraging powerful tools like Agents SDK, CrewAI, LangGraph, AutoGen, MCP, I deliver robust prototypes, fast. My engineering-first approach means you get clear solutions, not confusing black boxes. Based in Vienna, my goal is simple: to make your business more efficient and more profitable.
From Hakkari to Vienna: A journey of passion, innovation, and AI excellence
I'm a passionate AI Engineer specializing in building autonomous agent systems that revolutionize how businesses operate. With deep expertise in OpenAI Agents SDK, CrewAI, LangChain, LangGraph, AutoGen and MCP technologies, I create intelligent solutions that automate complex workflows and drive real business value.
Born in the mountains of Hakkari, Turkey, I discovered my passion for technology early. Now thriving in Vienna's dynamic tech ecosystem, I bridge the gap between cutting-edge AI research and practical business applications. My multicultural background gives me unique insights into solving diverse technical challenges.
LangChain, LangGraph, CrewAI, AutoGen, Multi-agent orchestration
Python, FastAPI, Docker, API Development
Deep Learning, NLP, Computer Vision, MLOps
Building interactive UIs for AI applications with Gradio, Streamlit, and foundational web technologies (JavaScript, HTML5, CSS).
Business Analysis, Problem Analysis, AI Agent Design, System Integration
Comprehensive AI engineering services designed to transform your business operations
Design and build bespoke AI agent crews using LangChain, CrewAI, LangGraph and AutoGen frameworks. From concept to deployment, I create intelligent systems that work autonomously and efficiently.
Modern, responsive websites with integrated AI capabilities. Combine beautiful design with intelligent features like chatbots, recommendation systems, and automated content generation.
Strategic guidance to identify AI opportunities, create implementation roadmaps, and ensure successful AI adoption. Perfect for businesses ready to embrace AI transformation.
Real-world applications of advanced AI technologies solving complex business challenges
Fine-tuned a Llama 3.1 model using the Q-LoRA technique to predict product prices from text descriptions. The resulting specialized model achieved a 70% success rate, significantly outperforming powerful proprietary models like GPT-4o on this specific regression task.
Developed a complete Retrieval-Augmented Generation (RAG) system for a knowledge base. The project involved manual implementation to grasp core concepts, followed by using LangChain, ChromaDB, and FAISS for efficient, accurate, and scalable Q&A.
Built a simulated software engineering team using crewAI to automate the development lifecycle. The crew, configured via YAML, autonomously handles tasks from technical design and coding to unit testing within a secure Docker environment.
Constructed a stateful agent using LangGraph capable of cyclical, self-correcting behavior. The agent evaluates its own output, refines it based on feedback in a loop, and maintains conversation history using persistent checkpointing with SQLite.
Designed and orchestrated a team of four specialized agents (planner, searcher, writer, emailer) using the OpenAI Agents SDK. This system automates complex research tasks from planning and web searches to report generation and final delivery via email.
Developed a multi-modal AI assistant for an airline's customer service. This chatbot can not only process and respond with text but also generate and deliver answers in voice and image formats, creating a richer user experience.
How a specialized Llama 3.1 model outperformed GPT-4o on a complex regression task.
General-purpose models like GPT-4o often struggle with highly specific, niche tasks like predicting product prices from unstructured text descriptions. The goal was to create a specialized model that could accurately perform this regression task, delivering precise and reliable price predictions where generic models failed.
I selected Llama 3.1 for its strong baseline performance and fine-tuned it using the memory-efficient Q-LoRA technique. The process involved meticulous data preparation, hyperparameter tuning, and rigorous evaluation using Weights & Biases for tracking experiments. This engineering-first approach ensured that every step was optimized for the final outcome.
Achieved an average error of just $46, proving the model's precision in a difficult prediction task.
The fine-tuned model significantly outperformed the much larger and more expensive GPT-4o, showing the power of specialization.
With a 70% success rate on predictions, the model demonstrated its reliability and business value for automated pricing tasks.
Stay updated with the latest AI trends, tutorials, and industry insights
A comprehensive guide to creating intelligent agent crews that collaborate to solve complex problems. Learn the fundamentals and best practices.
Read More →Exploring how autonomous AI agents are transforming industries and what businesses need to know to stay competitive.
Read More →A detailed comparison of popular AI frameworks to help you choose the best tool for your specific use case and requirements.
Read More →Ready to transform your business with AI? Let's discuss your project and create something extraordinary together.
Whether you need a custom AI solution, strategic guidance, or just want to explore the possibilities of AI for your business, I'm here to help.