Book this talent
Book this talent
Middle
Uzbekistan
Results-driven AI/ML Engineer with experience building and deploying scalable LLM and machine learning systems across research and production environments. Skilled in RAG architectures, LLM/VLM inference pipelines, MLOps, and end-to-end ML development using Python, PyTorch, FastAPI, Docker, and SQL. Proven track record of improving model performance, scaling AI products to real users, and delivering production-ready solutions in international teams across Germany and Uzbekistan. Experienced in observability and experimentation tools including Langfuse and Weights & Biases, with strong foundations in data engineering, deep learning, and software development. First-class Business Information Systems graduate and DAAD scholarship recipient with strong problem-solving skills demonstrated through competitive ML and algorithmic achievements.
Technical University of Berlin
Westminster International University in Tashkent
• Engineered and scaled a RAG-based customer support chatbot to 200+ daily users, utilizing n8n for orchestration, Qdrant for vector storage, and Langfuse for observability and performance tracking.
• Mentored junior AI Engineers on best practices for LLM deployment and observability, increasing team velocity and ensuring high standards for code maintainability.
• Collaborated with cross-functional stakeholders to define technical specifications for AI initiatives, ensuring an alignment between business objectives and engineering execution.
• Processed custom, noisy real-world datasets using SQL for querying and transformation, combined with data cleaning, labeling, and feature extraction, preparing datasets for machine learning pipelines, including training, evaluation, and deployment.
• Trained classical machine learning models (e.g., Random Forest, XGBoost, CatBoost) as well as deep learning models with PyTorch, achieving 30% higher accuracy compared to the baseline models.
• Integrated Weights, & Biases (W&B) for experiment tracking, performance visualization. As well as Docker and FastAPI for model inference.
• Designed and deployed a containerized inference pipeline for LLMs using FastAPI, Docker, and HuggingFace Transformers, enabling real-time performance on visual reasoning tasks. Track progress with Weights, & Biases and trained models on multiple GPUs.
• Worked extensively with LLMs and vision-language models (VLMs) to build AI systems capable of interpreting and reasoning about complex multimodal data.
• Built a complete ML pipeline from data loading and preprocessing to training and LLM inference on a server, leveraging NumPy, Pandas, and advanced SQL for data wrangling, used Matplotlib and Seaborn for visualization, and PyTorch for model development.
• Grew the number of active users to 5,000 by implementing new features and improving app performance based on user feedback and analytics
• Collaborated with cross-functional teams developers, designers, and stakeholders to ensure alignment with business objectives and successful project delivery.
• Led end-to-end projects from ideation stage to the production, identifying essential features and aligning development with user needs and business strategy.
Talent Pool
