Machine Learning & Python Developer
Passionate about building practical AI tools and bots that actually work ๐
Telegram bot that detects, extracts, and classifies text from document images. (17 languages)
Built using MMOCR, Tesseract, and FastText.
๐ View Project
๐ Also published related dataset:
Python library that converts images and videos to ASCII art.
Supports real-time processing with NumPy and OpenCV.
๐ View Project
๐๏ธ Also check out the CLI video player based on it:
๐ฎ ascii-video-player
Graduation project using Saiga LLaMA-3 8B fine-tuned on hh.ru to generate job listings.
Includes Dockerized Flask backend and React frontend.
๐ View Backend ๐ View Frontend
๐ Trained and compared multiple models, collected into:
Simulated grocery scale that uses YOLOv8 to detect 20+ products.
Integrated with PostgreSQL and served via Docker.
๐ View application ๐ View docker
- ๐ docs_on_several_languages โ multilingual OCR dataset
- ๐ vacancy-collection โ LLM models & datasets for job generation
- Testing different preprocessing methods (Sobel, FFT, Gabor, etc.) to improve small object detection models (YOLOv8n)
- Email:
timoshin_aleksey02@mail.ru
- Telegram: @alekseyscorpi
- GitHub: @AlekseyScorpi
- Huggingface: AlekseyScorpi (Aleksey Timoshin)