sajad ghawami

Writing on systems, retrieval, calibration, and the architecture of honest arguments. Long-form essays that observe and reason — rarely prescriptive, always patient.

Good Rankings, Wrong Probabilities: A Calibration Audit of Multimodal Cancer Survival ModelsRoad to GraphRAG: Vector Search Alone Can't Ground LLMsBridging LLMs and Databases: A Scalable Approach to Natural Language Querying of Tabular Data from CSV, Excel and Beyond