Prasanna Karjagi

Passionate engineer working on AI, computer vision with practical examples and eagerly waiting to explore more which helps me to gain knowledge on the new AI Aura.

• Received a customer excellence award for outstanding support and service delivery.
• Developed and deployed AI applications onsite in Europe, specifically in Italy and Austria.
• I have hands-on experience in developing and deploying AI applications and conducting Proof of concepts in different European countries.

• Analyzed client needs and created detailed feasibility reports.
• Developed Proof of Concepts (POCs) to validate new software applications before full-scale development
• Engineered comprehensive solutions encompassing model development, backend integration, and application dockerization.
• Model Development: Conducted data gathering and cleaning, selected optimal algorithms, and trained models using TensorFlow and PyTorch and Opencv.
• Backend Integration: Established API endpoints to handle image data, process it with models, and deliver results.
• Docker of application: Designed Dockerfiles specifying dependencies, configurations, and code to facilitate seamless deployment and scalability.
• Deployment: Implemented applications on servers or cloud platforms by configuring servers, installing necessary dependencies, and deploying Docker containers
• Installed and commissioned vision applications in Original Equipment Manufacturer environments, ensuring proper setup and functionality.
• Installation: Setting up hardware and software components of the vision system, including cameras, light, PLC's, and computing units.
• Calibration: Fine-tuning the vision system to ensure accurate image capture and processing, adjusting parameters for optimal performance.
• Integration: Seamlessly integrating the vision system with existing manufacturing processes and equipment, ensuring smooth operation and data flow.
• Testing and Validation: Conducting thorough testing to validate the performance of the vision system, making necessary adjustments to meet OEM standards.
• Training and Support: Providing training to OEM staff on system operation and maintenance, and offering ongoing technical support to address any issues.

• Cable Classification, Developed and deployed a cable classification model with 99.8% accuracy using Python Tkinter for user-friendly operation.
• Detzo Coil Inspection Project, Implemented and deployed a robust vision application for Detzo coil inspection, which includes dot matrix decoding, OCR reading, and various solder defect detection to enhance quality control and manufacturing efficiency to a lightweight real-time anomaly detection system in one of the major leading OEM.
• Engine Inspection, Developed and implemented a visual inspection system for pre-delivery engine inspections, integrating component presence detection, OCR for data matching with databases, utilizing multiple cameras and Android mobile devices for comprehensive inspection capabilities.
• Front Floor Stud Inspection Project, Developed and implemented a comprehensive automated visual inspection system for front floor studs at a leading automotive OEM. The system is designed to check for the presence or absence of studs, accurately count them, verify their positioning within specified boundaries with a 2mm accuracy, and identify broken studs. This ensures that only intact studs proceed in the manufacturing process.
• Hemlock Semiconductor (HSC) Project for Automated Inspection of Bag Labels, Developed and implemented an automated inspection system for Hemlock Semiconductor (HSC), involving segmentation models, object detection models, and OCR libraries to assess label placement, orientation, condition, barcode quality, label information, and defect detection in bags.
• Labels Inspection on UPS, Implemented a mobile inspection application deployed on the cloud at Schneider Electric Philippines CAV3 plant to ensure UPS labels correspond accurately with their model numbers.
• Pad Printing Inspection, Developed a solution to identify defects in real-time, preventing production issues and reducing wastage by inspecting thumb prints, scratches, dust bubbles, and print shifts.