Ruany Doehnert

Part of the team delivering a payroll reconciliation project for a large organization, navigating data challenges to improve accuracy and efficiency. Iterated on predictive models using Python to debug inconsistencies in HR/payroll datasets, reducing errors by 25% through supervised learning, cross-validation, and experimental plans to identify root causes. Designed and evaluated benchmarks by analyzing datasets, building ETL scripts in Microsoft Fabric, and providing structured feedback to optimize workflows for better model performance. Communicated findings via Power BI dashboards and presentations, showing work through reports on model iterations and dataset improvements to support AI needs.

Solved applied ML tasks by iterating on supervised/unsupervised models and reinforcement learning algorithms in Azure, optimizing for pricing and forecasting. Led POC projects with Azure OpenAI LLMs, debugging NLP-related behaviours and documenting failure modes for iterative improvements. Designed benchmark tasks, built evaluation scripts for real-time data integration in mining safety models, and used experimental plans to determine cause-and-effect relationships. Improved training datasets through structured feedback on feature selection and model evaluation; presented insights via data visualizations and collaborated with cross-functional teams for production-ready solutions. Consulted with customers on AI & GenAI needs, structuring requirements and selecting candidate ML services and algorithms. Achievements: Mining Safety AI and price optimization example outcomes.

Iterated on supervised learning models using Python (pandas, scikit-learn), debugging performance with RMSE and cross-validation. Designed optimization benchmarks, analyzed datasets, applied probabilistic modelling, and built PuLP-based scripts for resource allocation. Evaluated model behaviours in NLP contexts (forecasting trends), documenting iterations and communicating results via stakeholder reports. Key achievement: an optimization model that reduced transportation costs by 20% and improved delivery efficiency by 15%.