Дмитро

DMYTRO SANZHARIVSKYI
AI/ML ENGINEER
[відкрити контакти](див. вище в блоці «контактна інформація») [відкрити контакти](див. вище в блоці «контактна інформація»)
https://github.com/wroole

SUMMARY
Machine Learning and AI-focused Electrical Engineering & Informatics student with hands-
on experience building ML and LLM-powered applications. Experienced with Python,
PyTorch, Scikit-learn, FastAPI, Docker, and AWS, as well as modern AI technologies
including RAG, LangChain, vector databases, Hugging Face, and LoRA fine- tuning.
I have experience developing AI assistants, integrating LLMs into real-world applications,
and building end-to-end systems from data processing and model development to API
integration and deployment. Comfortable working with CLI-based AI development tools and
agents, including Claude, Codex, and Antigravity.

EDUCATION
Bachelor’s Degree in Electrical Engineering & Informatics
Technical University of Košice (TUKE) | 3rd year 2024 – 2027

HACKATHONS
Erste Digital Hackathon (AI Shopping Assistant) november 2025
Developed an AI-powered shopping assistant using OpenAI
API and a locally fine-tuned LLaMA model.
Fine-tuned the LLM with LoRA for natural-language-to-SQL
query generation.
Built a FastAPI backend for real-time inference and data
processing.
Implemented dynamic expense analysis and visualization
using Python and Matplotlib.

Telekom Slovakia hackathon (AI Planning Assistant) november 2025
Built an AI-powered assistant for intelligent daily planning
and task optimization using the OpenAI API.
Added voice control to enable hands-free interaction with
the assistant.
Integrated external context such as weather data to
generate more personalized recommendations.
Developed a FastAPI-based REST backend to handle
planning logic and AI interactions.

KEY COMPETENCIES
Python, SQL
Git, Linux, AWS (EC2), PostgreSQL
NumPy, Pandas, Matplotlib, Seaborn
Scikit-learn, PyTorch
Hugging Face, LLM Fine-tuning (LoRA)
RAG, LangChain, LangGraph, Vector Databases (Qdrant, ChromaDB), n8n
REST APIs, FAST API Docker, Docker compose, MLOps, GitLab CI/CD
AI Agents, CLI-based AI Development (Claude, Codex, Antigravity)

PROJECTS
AI CV Assistant
Tech Stack: Python, FastAPI, LangChain,
LangGraph, Qdrant, n8n, PostgreSQL, React,
Ollama, OpenAI API, SQLAlchemy, Docker

Modular AI Pipelines: Built an end-to-end CV evaluation system using LangChain with
9 parallel chains (ATS scoring, grammar, skills extraction, structural review, and bullet
rewriting).
RAG Search Agent: Developed an autonomous candidate search agent using
LangGraph and Qdrant Vector DB with automated query rewriting and document
grading.
AI Automation & Integrations: Built a lightweight n8n workflow for event-driven
candidate evaluation, routing results based on score thresholds and sending AI-
generated candidate summaries to Telegram or Discord.
Security & Processing: Implemented heuristic Prompt Injection defense, PDF
extraction via PyMuPDF, and deterministic feature extraction.
Engineered a REST backend with FastAPI, PostgreSQL, SQLAlchemy, and JWT
authentication; containerized the multi-service architecture using Docker & Docker
Compose.

Bank Customer Churn Prediction
Tech Stack: Python, Pandas, NumPy, Scikit-learn,
XGBoost, Joblib, FastAPI, Pydantic, Streamlit,
Matplotlib, Seaborn, Docker & Docker Compose

Performed exploratory data analysis to identify patterns and factors associated with
customer churn.
Engineered features and compared multiple machine learning models to select the
best-performing approach.
Deployed the final model as part of a multi-service architecture using Docker
Compose.