AI-Powered Part Identification & Crew Allocation
Company: IBM
Role: Data Scientist
📌Project overview
Delivered two enterprise-grade AI solutions in the automotive and logistics sectors for Daimler Truck and Detroit Diesel Corporation.
❗The problem
Misidentification of engine parts was causing high rejection rates; separately, inefficient crew scheduling was increasing customer downtime.
🛠️ What I did
- Trained object detection models on isometric images for automated part classification (PyTorch + Blender API)
- Built a proximity-based auto-allocation system to optimize service response using Python + Azure ML
- Used KNNs and scheduling logic to recommend optimal crew assignments
- Integrated models into dashboards with Streamlit and ReactJS
🎯 Impact
- Reduced part rejection costs and improved production quality
- Saved 25–40 minutes of outage per day per customer via crew optimization