At General Motors, I designed and built JVL-TDP, a tightening-data platform that replaced ToolsNet 8 and delivered US$280K of immediate impact. It became a company-wide reference for standardizing tightening-data acquisition software.
For JVL-TDP, I built the ingestion layer for 280+ Bosch Rexroth controllers by interpreting captured TCP traffic, and reverse-engineered the proprietary BS350 binary protocol. I shipped six milestones through autonomous on-premise deployment, with observability in Prometheus, Grafana, and Loki.
On JVL-MIP, I built an ML pipeline that predicts torque at R²=0.9988 and flags thermal runaway and tool wear in an air-gapped environment. Across the Q-DAS ecosystem, I built a Flask portal, a real-time monitor for 60+ stations, and two read-only MCP servers for audited access to quality data.
My work at General Motors also includes quality engineering and manufacturing launches. I automated Q-DAS reporting, helped prevent cylinder-head scrap valued at about R$190K, and implemented an activity-management system adopted across South America Manufacturing Engineering.
