At Apexon, I built and optimized distributed data-processing pipelines handling 112,000+ clinical records using PySpark, SQL, and Azure Databricks. I also developed ETL workflows for healthcare data ingestion, transformation, validation, and storage.
I built OmniContext, a local-first RAG platform that turns repositories, PDFs, web content, and clipboard activity into persistent semantic memory. It combines semantic and keyword search with ChromaDB, SentenceTransformers, and Ollama for local retrieval and inference.
For my ICU Mortality Risk Prediction System, I developed an end-to-end MLOps pipeline and an XGBoost model that achieved 0.91 ROC-AUC. I implemented MLflow experiment tracking and model versioning, and built FastAPI inference APIs with Streamlit visualizations.
