Projects

A C# ASP.NET Core REST API for personal finance tracking, built around accounts, transactions, and transfers. It uses Clean Architecture, EF Core, PostgreSQL, JWT authentication, rate limiting, pagination, Docker, Swagger, and integration tests with xUnit and WebApplicationFactory. Deployed to Fly.io with Neon PostgreSQL.

E-commerce API learning project built while practicing ASP.NET Core. It covers product catalog, basket, identity, and checkout flows with an Angular frontend. Main goal was learning common backend patterns, not shipping a production product.

Windows desktop app in C#/.NET for tracking personal and family expenses. Built with WinForms, SQLite, full CRUD screens, and a repository pattern for data access. I used it personally for several months, which helped me practice practical desktop development, persistence, and organizing business logic in a small C# application.

Internal operations tool built with Next.js, TypeScript, PostgreSQL, and Prisma for real campsite workflows. It supports pitch availability search, role-based permissions, audit logging, web push notifications, and a reception waiting queue. Used in a real workplace, it helped me practice production-minded full-stack development.

This portfolio is a Next.js application with bilingual content, an admin panel, analytics, database-backed project management, and an EchoMind chatbot integration. It is where I practice shipping a polished public product, not just a static page: content editing, authentication, deployment, monitoring, and real visitor interactions.

Next.js frontend for the Finance Tracker API, built to practice connecting a production backend to a usable web interface. It provides CRUD screens for accounts and transactions, consumes the authenticated REST API, and makes the C# backend easier to demo and test from a browser.

n8n workflow for drafting Facebook posts with RAG context and generating a matching image. A Telegram approval step lets a human review, reject, or regenerate the content before anything is posted to the page.

n8n workflow for generating YouTube Shorts ideas, creating video drafts with Google Veo 3.1, and preparing them for publishing. It includes optional human review, retry handling, model selection, and Google Sheets tracking so each run can be checked instead of posted blindly.

Telegram-based assistant built with n8n to connect task, email, and calendar actions in one workflow. It accepts voice or text requests, keeps basic conversation context, and routes the request to the right service. Main goal was learning how to wire practical automations together.

RAG workflow built in n8n that syncs Google Drive documents into Pinecone and lets an assistant answer questions from that document set. I focused on document sync, retrieval, conversational memory, and keeping answers tied to the provided company files.

Built a small license plate recognition project with YOLOv8 for detection, EasyOCR for reading text, and OpenCV for image processing. The goal was to practice a full computer vision pipeline: detect the plate, crop the image, run OCR, and show the result clearly.

Security monitoring requires constant attention to webcam feeds, missing motion events when not actively watching. Built motion detection system using OpenCV that captures frames during motion events and sends batched email alerts with representative images every 15 seconds. Automated security monitoring with email notifications eliminates need for constant webcam watching.