Vlad

Yanhirov

Personal AI Telegram Bot (LLM + Voice + Vision) Built a highly optimized, context-aware AI companion capable of processing text, voice, and images in real-time. • AI & API Integration: Implemented low-latency LLM and Vision inference via Groq API, coupled with local zero-shot Voice Cloning using Coqui XTTS. • Hardware Optimization: Profiled application memory usage to run heavy machine learning models efficiently on constrained hardware (15GB T4 GPU), implementing memory management techniques to prevent Out-Of-Memory (OOM) errors. • Bot Engineering: Developed a robust asynchronous Telegram bot interface using Aiogram to handle multi-format user inputs continuously.

Web Telegram Shop (TMA + Crypto Payments) Designed and deployed a production-ready e-commerce backend integrated directly into Telegram Mini Apps. • Architecture: Architected the system strictly decoupling business logic from frameworks (Use Cases & Repositories) for high maintainability and testability. • FinTech Integration: Engineered secure, asynchronous Webhook handlers for cryptocurrency gateways (NOWPayments, CryptoBot) to process global transactions with zero chargebacks. • Infrastructure: Fully containerized the application stack using Docker and Docker Compose, optimizing deployment for cloud environments. • Automation: Set up continuous integration pipelines (GitHub Actions) for automated QA checks (Ruff, Mypy, Pytest) and configured Celery/Redis for background task processing.