I built a Predictive Maintenance AI System that predicts five types of machine failure from industrial sensor data. Its XGBoost model reached 99% accuracy and a 0.96 ROC-AUC, and I turned it into a Streamlit app for real-time predictions.
For my Enterprise RAG Document Assistant, I built a pipeline that answers questions from company PDFs using document content. It retrieves relevant passages and generates answers with LLaMA-3.3-70B, helping avoid unsupported responses.
I packaged the RAG app with Docker Compose, separating its backend and frontend for one-command setup. I also resolved a PyTorch version conflict that had interrupted the embedding step.
During my Advanced Data Science training, I cleaned and explored datasets in Python and created charts and model evaluation reports. I completed a B.Tech in Computer Science (AI/ML) at GLA University.
