These projects were completed as coursework in the UT Austin Post-Graduate Program in AI and Machine Learning for Business Applications. They are shown as examples of working methods, not as client engagements.

ReneWind: Predictive Maintenance

Neural-network classification of wind turbine failures from sensor data, tuned to minimize missed failures.

Medical Assistant: Document Q&A

A retrieval-augmented generation (RAG) tool over medical manuals, using a language model and prompt engineering.

HelmNet: Safety Compliance Detection

A convolutional neural network that detects whether workers are wearing safety helmets, using transfer learning and data augmentation.

SuperKart: Forecasting and Deployment

A sales forecasting model deployed as an API with Docker and Flask, with a Streamlit front end and Hugging Face hosting.

EasyVisa: Approval Prediction

Bagging, boosting, and stacking models that predict application outcomes and identify the factors that matter most.

The full verified portfolio is at mygreatlearning.com/eportfolio/peter-drozd.