At HARMAN International, I contribute to a Python and Flask-based CRS Intake Tool that turns unstructured customer documents into structured requirement data for IBM DNG. I helped reduce manual requirement-intake effort by approximately 60% while validating the workflow with customer documents.
I designed and implemented hybrid document processing that uses DocLayout-YOLO to assess page layouts, routes simpler pages to native Python parsers, and uses Docling for complex or mixed layouts. I also added extraction checks and fallback routing for failed or low-quality results.
In my earlier HARMAN role, I analyzed customer specifications and translated them into system-level requirements for automotive HMI and ECU functions. I led a five-member HMI requirements team and worked on requirements covering CAN communication, UDS diagnostics, and ECU interfaces.
I also develop Python utilities for requirement comparison, duplicate detection, and validation reporting. My systems engineering background includes automotive feature modeling with SysML/MBSE and experience with IBM DNG and Enterprise Architect.
