At Volvo, I built a platform that processes raw vehicle CAN logs, decodes them with DBC files, and separates signals for downstream teams. Automating the mapping between software releases and DBC versions helped ensure files decode correctly as vehicle software changes.
I orchestrate CAN data extraction with Airflow, S3, Spark, and Iceberg across more than 4,000 parameters. This reduced analysts’ data preparation time by more than 50%, and a Spark data quality framework with Grafana dashboards surfaces anomalies early.
I also built Databricks pipelines and a daily emulation process to compare reconstructed charging sessions with vehicle-aggregated sessions, supporting regression testing before production. A FastAPI charging data product joins validated session data with vehicle faults, while an Iceberg-based catalog supports teams building GenAI charging agents.
At Accenture, I developed HAdeaP, a patented healthcare ETL application, and co-developed the patented Digital Data Assistant using natural language processing and deep learning. I also replaced an Informatica workflow with a Python ETL framework that improved performance by 30%.
