Desktop Support Technician

 

Working as Desktop Support at Geek Squad was a great way for me to combine my love of hardware and software repairs with a job that would enable my next steps in my IT career. As a Desktop Support Technician (known internally as Advanced Repair Agent) I was the second level of support. I was in charge of performing advanced repairs and upgrades to the hardware and operating systems of laptops, desktops, and phones. 

 

Tools learned:


A large problem at our Geek Squad location was response times for people in line. For privacy and data protections purposes, most of the workbench was not in view of the public. There were walls positioned strategically. We had a camera pointed at the line area and it displayed out to a display we had, but it was also placed in an inconvenient spot. When you are performing detailed repairs on an iPhone or are surrounded by screws as you try to reinstall a desktop motherboard, it is easy to get tunnel vision and not see the person in line. The rule was to look up every 5 minutes, but no one is perfect at time keeping and even 5 minutes can be a long time to be in line with no signs of anyone around. 

Since I had become interested in the raspberry pi and seeing just how much you can accomplish with a small Linux device, some hard work, and loads of documentation, I came up with an idea to address this problem. After seeing videos of people using cameras with raspberry pis as make shift dash cams with object detection, or home surveillance with object detection, I was confident that I could leverage this as a solution. Using TensorFlow Lite on a raspberry pi that I convinced my manager to buy, I connected the camera we already had pointing at the line to the device. It was actually a lot easier than it sounds. The models are already trained and available to the public, but you can make your own. I set up a custom zone (box) around the bollards where people wait in line and I programmed it to look for a person standing in that zone for more than 10 seconds. This was to remove the false positives of people walking by and looking in the shelves. I had an audio clip using Text-to-speech that said "Person in line". Whenever Tensorflow met all the parameters for a person being in line, it would play the audio.

After a lot of trial and error and major tweaking, this project was successful. We were able to cut down our response time from 5 minutes to about 15 seconds. This meant we could focus on our work without being paranoid of people being in line. And also when people actually were in line, we could give them great and speedy service. First impressions are critical, and this allowed ours to be much better for clients.

One of the most tedious tasks was taking inventory of our parts for mobile repairs for iPhones and android phones. We had problems with the packages being sent to us correctly as well as our shipments of mobile parts back to their vendors making it there. Someone years prior had created a spreadsheet that was to be filled out every time we received a part and every time we shipped out a part. This included tracking number, serial number, part number, and a lot of other alpha-numeric codes which were tedious to write. The spreadsheet was very small, and it was easy to confuse a G for a 6 or an A for a 4. Everyone hated the process but it had to be done. I was sure we could find a better way, so I started drafting some ideas.

Every alpha-numeric code that we were writing down also had a barcode next to it. Everyone doing inventory, like the package couriers and the parts vendors use the barcodes for quick and efficient inventory taking. So why didn't we use it? So I created a fairly simple spreadsheet that aided the whole process. The first field is your name which you fill out. Then you push tab and scan the series of barcodes one after another. Suddenly, this act of documenting every part being shipped went from 60 seconds to 5-10 seconds. Furthermore, it also meant that there were no typos. No problems with legibility. And the icing on the cake was that we could search through the spreadsheet with ease using the Find tool. We would save 100s of dollars per month because this cut down on inventory errors and allowed us to have the accountability that we needed.