Master's student at HAW Kiel with 2+ years production experience.
Building intelligent systems with
LangGraph,
FastAPI, and
Docker.
I'm a Master's student in Data Science at HAW Kiel with 2+ years of production AI experience. I bring a unique combination of academic excellence and real-world engineering skills.
Built and deployed chatbots at Bank of Georgia, improving accuracy by 15% using modern NLU and RAG techniques.
Currently building unimate - a LangGraph-based multi-agent system for German exam preparation with RAG architecture.
Experienced with Docker, FastAPI, AWS (SageMaker, S3, EC2), and CI/CD pipelines for scalable ML deployments.
German B2 (speaking), English C1. Experience working in international teams across Georgia, Poland, and Germany.
I've shipped real AI products that serve thousands of users daily
Production-ready skills in modern AI/ML technologies
Building multi-agent systems
Retrieval-Augmented Generation
Production APIs
Containerization
B2 (Speaking)
C1 (Professional)
In Progress - Feb 2025
Completed - 2021
Specialization - 2025
Production-ready AI systems you can try right now
Building a cutting-edge multi-agent system using LangGraph and RAG architecture for TestDaF/Goethe exam preparation. This production-ready system demonstrates my expertise in modern AI engineering.
from langgraph.graph import StateGraph
from fastapi import FastAPI
import docker
# Multi-agent system for German exam prep
class UnimateAgent:
def __init__(self):
self.graph = StateGraph()
self.rag_engine = RAGEngine()
async def process_query(self, query: str):
# Intelligent routing to specialized agents
return await self.graph.run(query)
Production-ready RAG chatbot using Gemini API for intelligent document Q&A. Demonstrates real-world implementation of retrieval-augmented generation.
Scalable ML API with 90% accuracy using XGBoost. Fully containerized with Docker for easy deployment. Production-ready with proper error handling and logging.
PyTorch CNN for pneumonia detection served via FastAPI. Production-ready endpoints with proper validation and Docker deployment.
ML model achieving 85% accuracy in predicting telecom customer churn. Includes comprehensive EDA and feature engineering.
MediaLab 5K Lari Winner (2024). Converts blog content to animated videos using Stable Diffusion and NLP.
Named Entity Recognition model fine-tuned on CoNLL-2003 dataset using BERT and PyTorch. Demonstrates advanced NLP capabilities.
Feb 2024 - Feb 2026 (Expected)
2019 - 2024
Looking for a Werkstudent (20h/week) or internship position in AI
Engineering / Data Engineering.
Available for hybrid work in Greifswald/Berlin area.
I respond within 24 hours • Available for video calls
Phone
+49 152 1023 3165Location
Greifswald / Berlin, Germany