Hubert User

Activities as contractor:
- Implemented dynamic clustering and embedding analysis using AWS Bedrock (Titan models) and Claude for automatic topic labeling and description generation.
- Enhanced insight accuracy and explainability of internal data products through automated topic discovery and sentiment correlation.
- Designed and maintained production pipelines in Airflow/Kubernetes, integrated

- Designed and delivered enterprise-scale LLM applications
leveraging Azure OpenAI, handling hundreds of concurrent
requests per minute.

- Built a prompt optimization platform using multi-armed bandit
algorithms and LLM evaluation metrics (BLEU, ROUGE, semantic
similarity) to iteratively improve prompt quality based on user
feedback.

- Developed an AI-driven assessment generation system that
red

- Architected and deployed an AI-powered retail analytics chatbot using Azure AI Studio and GPT-4 technology.
- Implemented end-to-end solution leveraging PromptFlow for advanced prompt engineering and LLM orchestration.
- Developed custom conversational flows enabling store managers to obtain instant insights from retail data.
- Successfully delivered POC demonstrating significant potential for

Worked on developing a predictive revenue model for mobile games. Responsibilities:
- handling large datasets that required preprocessing and modeling, utilizing tools such as BigQuery, XGBoost, and MLflow
- creating and evaluating the revenue prediction models
- maintaining machine learning pipelines

- developed and deployed a license-plate reader application (computer vision based) with over 85% accuracy.
- led a recommendation service project, improved top 1 accuracy by 12% for the deep learning based recommendation service.
- created an auto retraining system for deep learning models on the scale of tens of thousands of models.
- implemented a Spring service which is able to smartly correct

- development of a system for borrowing books from a library (training project)
- processing raw analytical data and applying it for machine learning (training project).

Tech stack: Python, Git, Flask, MySQL, MongoDB, NumPy, Pandas, TensorFlow, Docker, Bash.

Completed undergraduate studies in Applied Computer Science, building a strong foundation in computer science and data analysis.