Projects
Adapting LLMs to Uzbek Language
This project focuses on adapting large language models such as Mistral-7B, Mistral-12B, and LLaMA-8B to the Uzbek language. The goal is to improve natural language processing capabilities for Uzbek, enhancing applications in machine translation, text generation, and language understanding.
- Implemented fine-tuning techniques for large language models
- Developed a custom dataset of Uzbek text for training
- Evaluated model performance using various metrics
- Published preliminary results demonstrating improved performance on Uzbek language tasks
Cartouch Backend Development
As a Machine Learning Engineer at Cartouch, I developed backend systems for deploying machine learning models. This project involved creating robust and scalable infrastructure to support various computer vision applications.
- Utilized Django and FastAPI for creating RESTful APIs
- Implemented efficient data processing pipelines
- Integrated machine learning models into the backend architecture
- Optimized system performance for handling large-scale requests