Barcode Scanning:

• Integrated barcode scanning capability for part numbers and locations, enabling faster and more accurate part identification and updates.

Optical Character Recognition (OCR):

• Implemented OCR to read pick lists and generate sorted location lists automatically for multiple parts, streamlining the picking process.

Android App Integration:

• Adapted the tool into an Android app that runs on Zebra Z-21 Mobile Computers, providing mobile access to the part location system.

Real-Time Location Updates:

• The app allows material handlers to update part locations instantly by scanning location barcodes after moving parts, ensuring location data remains accurate.

Google Drive Integration:

• Integrated with Google Drive to enable seamless, real-time updates to part location data, ensuring it is always accessible and up-to-date.

Shelf Location Map:

• Added a visual map highlighting shelf locations, making it easier for users to find parts in the warehouse.

During my summer internship at Veolia’s Manufacturing Facility in Minnetonka, I worked primarily in the area where they assembled Tonkaflo Pumps. This was an uncontrolled section of the warehouse, meaning there wasn’t a designated location for each part. Assemblers spent nearly half their time searching for parts, and material handlers frequently had to ask where specific parts should go when restocking. When I inquired about why this area hadn’t been converted to a controlled space, I was told that making the necessary changes in SAP would be too time-consuming and resource-intensive.

Recognizing the inefficiency, I proposed a solution that would allow both assemblers and material handlers to locate parts more quickly without needing any changes to SAP. I built a Python program with a simple GUI, enabling users to quickly search for parts and their locations. After creating a working prototype within a couple of hours, I presented it to my supervisor, who was interested in the idea and encouraged me to pursue it alongside my other tasks. Eventually, I was asked to focus solely on this project.

To ensure the tool would be truly effective, I spent time on the floor talking with both assemblers and material handlers. Their primary concern was that parts were frequently moved to different shelves, which made it difficult to maintain an accurate system. To address this, I developed an additional feature that allowed for rapid updates to part locations via a simple GUI. The material handlers also wanted the ability to scan barcodes instead of manually entering part numbers, so I repurposed an old barcode scanner from IT and integrated it into the system. The assemblers wanted the ability to input multiple parts at once and receive a pick list sorted by location, which led me to implement optical character recognition (OCR) to read pick lists and automatically generate sorted location lists.

With these requests in mind, I combined all the features into a single program and arranged a meeting with my supervisor, the assembler and material handler supervisors, as well as the assemblers and material handlers themselves. During the presentation, they were excited about the tool’s functionality, but they wanted to run the program on their Zebra Z-21 Mobile Computers, which operated on Android. Despite having no prior experience in app development, I agreed to adapt the program and used Veolia’s internal LLM and Android Studio to create an Android app with all the original functionality and more.

The Android app featured barcode scanning, OCR, and Google Drive integration, allowing for seamless updates to part location data. I also added a map highlighting shelf locations, making it even easier to find parts. One key improvement was the ability to scan a location barcode right after scanning a part number, allowing for quick updates when parts were moved to different shelves. This system made it easy to manage part locations and ensured that updates could be made in real time.

By the end of my internship, the app was actively being used by both assemblers and material handlers, reducing the time spent searching for parts and streamlining the restocking process by an estimated 20%. To ensure the system would continue to function smoothly after my departure, I developed a detailed work instruction. This document walked through how to modify any part of the app and explained how to leverage Veolia’s LLM to understand the underlying code. It provided clear guidance for future developers, enabling them to add new features or troubleshoot issues by using the LLM to interpret specific parts of the code.

The work instruction also included steps on integrating new functionalities or devices, such as adding compatibility with different barcode scanners or expanding the OCR system for other pick list formats. By outlining how to incorporate Veolia’s LLM into the development process, I ensured that future developers could not only maintain the app but also use AI to generate insights and solutions for further improvements.

In the end, I was able to create a tool that saved significant time in the assembly and restocking processes, while also setting up a framework for future development that didn’t require extensive knowledge of the codebase—just the ability to interact with the LLM.