>>C G technology is focused on data where we have long believed that data is in our DNA. We help maximize humanity's potential by delivering world class, precision engineered data solutions developed through sustainable and profitable partnerships. Included in our offerings are hard disk drives. As I'm sure many of you know, ah, hard drive consists of a slider also known as a drive head or transducer attached to a head gimbal assembly. I had stack assembly made up of multiple head gimbal assemblies and a drive enclosure with one or more platters, or just that the head stacked assembles into. And while the concept hasn't changed, hard drive technology has progressed well beyond the initial five megabytes, 500 quarter inch drives that Seagate first produced. And, I think 1983. We have just announced in 18 terabytes 3.5 inch drive with nine flatters on a single head stack assembly with dual head stack assemblies this calendar year, the complexity of these drives further than need to incorporate Edge analytics at operation sites, so G Edward stemming established the concept of continual improvement and everything that we do, especially in product development and operations and at the end of World War Two, he embarked on a mission with support from the US government to help Japan recover from its four time losses. He established the concept of continual improvement and statistical process control to the leaders of prominent organizations within Japan. And because of this, he was honored by the Japanese emperor with the second order of the sacred treasure for his teachings, the only non Japanese to receive this honor in hundreds of years. Japan's quality control is now world famous, as many of you may know, and based on my own experience and product development, it is clear that they made a major impact on Japan's recovery after the war at Sea Gate. The work that we've been doing and adopting new technologies has been our mantra at continual improvement. As part of this effort, we embarked on the adoption of new technologies in our global operations, which includes establishing machine learning and artificial intelligence at the edge and in doing so, continue to adopt our technical capabilities within data science and data engineering. >>So I'm a principal engineer and member of the Operations and Technology Advanced Analytics Group. We are a service organization for those organizations who need to make sense of the data that they have and in doing so, perhaps introduce a different way to create an analyzed new data. Making sense of the data that organizations have is a key aspect of the work that data scientist and engineers do. So I'm a project manager for an initiative adopting artificial intelligence methodologies for C Gate manufacturing, which is the reason why I'm talking to you today. I thought I'd start by first talking about what we do at Sea Gate and follow that with a brief on artificial intelligence and its role in manufacturing. And I'd like them to discuss how AI and machine Learning is being used at Sea Gate in developing Edge analytics, where Dr Enterprise and Cooper Netease automates deployment, scaling and management of container raised applications. So finally, I like to discuss where we are headed with this initiative and where Mirant is has a major role in case some of you are not conversant in machine learning, artificial intelligence and difference outside some definitions. To cite one source, machine learning is the scientific study of algorithms and statistical bottles without computer systems use to effectively perform a specific task without using explicit instructions, relying on patterns and inference Instead, thus, being seen as a subset of narrow artificial intelligence were analytics and decision making take place. The intent of machine learning is to use basic algorithms to perform different functions, such as classify images to type classified emails into spam and not spam, and predict weather. The idea and this is where the concept of narrow artificial intelligence comes in, is to make decisions of a preset type basically let a machine learn from itself. These types of machine learning includes supervised learning, unsupervised learning and reinforcement learning and in supervised learning. The system learns from previous examples that are provided, such as images of dogs that are labeled by type in unsupervised learning. The algorithms are left to themselves to find answers. For example, a Siris of images of dogs can be used to group them into categories by association that's color, length of coat, length of snout and so on. So in the last slide, I mentioned narrow a I a few times, and to explain it is common to describe in terms of two categories general and narrow or weak. So Many of us were first exposed to General Ai in popular science fiction movies like 2000 and One, A Space Odyssey and Terminator General Ai is a I that can successfully perform any intellectual task that a human can. And if you ask you Lawn Musk or Stephen Hawking, this is how they view the future with General Ai. If we're not careful on how it is implemented, so most of us hope that is more like this is friendly and helpful. Um, like Wally. The reality is that machines today are not only capable of weak or narrow, a I AI that is focused on a narrow, specific task like understanding, speech or finding objects and images. Alexa and Google Home are becoming very popular, and they can be found in many homes. Their narrow task is to recognize human speech and answer limited questions or perform simple tasks like raising the temperature in your home or ordering a pizza as long as you have already defined the order. Narrow. AI is also very useful for recognizing objects in images and even counting people as they go in and out of stores. As you can see in this example, so artificial intelligence supplies, machine learning analytics inference and other techniques which can be used to solve actual problems. The two examples here particle detection, an image anomaly detection have the potential to adopt edge analytics during the manufacturing process. Ah, common problem in clean rooms is spikes in particle count from particle detectors. With this application, we can provide context to particle events by monitoring the area around the machine and detecting when foreign objects like gloves enter areas where they should not. Image Anomaly detection historically has been accomplished at sea gate by operators in clean rooms, viewing each image one at a time for anomalies, creating models of various anomalies through machine learning. Methodologies can be used to run comparative analyses in a production environment where outliers can be detected through influence in an automated real Time analytics scenario. So anomaly detection is also frequently used in machine learning to find patterns or unusual events in our data. How do you know what you don't know? It's really what you ask, and the first step in anomaly detection is to use an algorithm to find patterns or relationships in your data. In this case, we're looking at hundreds of variables and finding relationships between them. We can then look at a subset of variables and determine how they are behaving in relation to each other. We use this baseline to define normal behavior and generate a model of it. In this case, we're building a model with three variables. We can then run this model against new data. Observations that do not fit in the model are defined as anomalies, and anomalies can be good or bad. It takes a subject matter expert to determine how to classify the anomalies on classify classification could be scrapped or okay to use. For example, the subject matter expert is assisting the machine to learn the rules. We then update the model with the classifications anomalies and start running again, and we can see that there are few that generate these models. Now. Secret factories generate hundreds of thousands of images every day. Many of these require human toe, look at them and make a decision. This is dull and steak prone work that is ideal for artificial intelligence. The initiative that I am project managing is intended to offer a solution that matches the continual increased complexity of the products we manufacture and that minimizes the need for manual inspection. The Edge Rx Smart manufacturing reference architecture er, is the initiative both how meat and I are working on and sorry to say that Hamid isn't here today. But as I said, you may have guessed. Our goal is to introduce early defect detection in every stage of our manufacturing process through a machine learning and real time analytics through inference. And in doing so, we will improve overall product quality, enjoy higher yields with lesser defects and produce higher Ma Jin's. Because this was entirely new. We established partnerships with H B within video and with Docker and Amaranthus two years ago to develop the capability that we now have as we deploy edge Rx to our operation sites in four continents from a hardware. Since H P. E. And in video has been an able partner in helping us develop an architecture that we have standardized on and on the software stack side doctor has been instrumental in helping us manage a very complex project with a steep learning curve for all concerned. To further clarify efforts to enable more a i N M l in factories. Theobald active was to determine an economical edge Compute that would access the latest AI NML technology using a standardized platform across all factories. This objective included providing an upgrade path that scales while minimizing disruption to existing factory systems and burden on factory information systems. Resource is the two parts to the compute solution are shown in the diagram, and the gateway device connects to see gates, existing factory information systems, architecture ER and does inference calculations. The second part is a training device for creating and updating models. All factories will need the Gateway device and the Compute Cluster on site, and to this day it remains to be seen if the training devices needed in other locations. But we do know that one devices capable of supporting multiple factories simultaneously there are also options for training on cloud based Resource is the stream storing appliance consists of a kubernetes cluster with GPU and CPU worker notes, as well as master notes and docker trusted registries. The GPU nodes are hardware based using H B E l 4000 edge lines, the balance our virtual machines and for machine learning. We've standardized on both the H B E. Apollo 6500 and the NVIDIA G X one, each with eight in video V 100 GP use. And, incidentally, the same technology enables augmented and virtual reality. Hardware is only one part of the equation. Our software stack consists of Docker Enterprise and Cooper Netease. As I mentioned previously, we've deployed these clusters at all of our operations sites with specific use. Case is planned for each site. Moran Tous has had a major impact on our ability to develop this capability by offering a stable platform in universal control plane that provides us, with the necessary metrics to determine the health of the Kubernetes cluster and the use of Dr Trusted Registry to maintain a secure repository for containers. And they have been an exceptional partner in our efforts to deploy clusters at multiple sites. At this point in our deployment efforts, we are on prem, but we are exploring cloud service options that include Miranda's next generation Docker enterprise offering that includes stack light in conjunction with multi cluster management. And to me, the concept of federation of multi cluster management is a requirement in our case because of the global nature of our business where our operation sites are on four continents. So Stack Light provides the hook of each cluster that banks multi cluster management and effective solution. Open source has been a major part of Project Athena, and there has been a debate about using Dr CE versus Dr Enterprise. And that decision was actually easy, given the advantages that Dr Enterprise would offer, especially during a nearly phase of development. Cooper Netease was a natural addition to the software stack and has been widely accepted. But we have also been a work to adopt such open source as rabbit and to messaging tensorflow and tensor rt, to name three good lab for developments and a number of others. As you see here, is well, and most of our programming programming has been in python. The results of our efforts so far have been excellent. We are seeing a six month return on investment from just one of seven clusters where the hardware and software cost approached close to $1 million. The performance on this cluster is now over three million images processed per day for their adoption has been growing, but the biggest challenge we've seen has been handling a steep learning curve. Installing and maintaining complex Cooper needs clusters in data centers that are not used to managing the unique aspect of clusters like this. And because of this, we have been considering adopting a control plane in the cloud with Kubernetes as the service supported by Miranda's. Even without considering, Kubernetes is a service. The concept of federation or multi cluster management has to be on her road map, especially considering the global nature of our company. Thank you.