Dr. Vasi Philomin from Amazon Cloud Technology: Seeing the world and the future from the forefront of generative AI

In 2023, generative artificial intelligence is rapidly becoming popular. Some high-tech talents who have been committed to the field of artificial intelligence seem to be pushed forward by this era, drawn into the torrent of history, and can no longer stop!

Global Vice President of Generative Artificial Intelligence at Amazon Cloud Technology, Dr. Vasi Philomin

The Path of Genius

One of our main strengths is bringing various technologies into the enterprise business environment and making them truly usable by users, including generative artificial intelligence. Dr. Vasi Philomin said that Amazon Cloud Technology is committed to generative artificial intelligence, not from scratch, but based on past accumulation and innovation

Take Amazon Cloud Technology as an example. As early as the 1990s, Amazon Cloud Technology was an initial attempt at human-computer interaction products. Today, the number of users of Amazon Cloud Technology is astonishing, with 100 million users and more than 1 billion interactions every week. From the perspective of actual application scenarios, Amazon Cloud Technology supports many human-machine collaboration scenarios. For example, in Amazon fulfillment centers, 1.6 million packages are sent every day. The company’s investment in infrastructure and supply chain management has made Amazon bigger and bigger in the retail business market

As the global vice president of generative AI at Amazon Cloud Technology, Dr. Vasi Philomin feels the same as most people. The development speed of generative AI is too fast, and the happiness of being recognized by the world comes too suddenly

If we go back to 1990, machine learning was still relatively unpopular at that time. However, Dr. Vasi Philomin was so passionate about it that he decided to make it his major. At the time, this choice he made was opposed by his family and friends, who thought he was crazy. However, Dr. Vasi Philomin eventually succeeded in obtaining a PhD in Computer Science from the University of Maryland, College Park, specializing in Computer Vision and Machine Learning, and also holds two Master's degrees in Mechanical Engineering and Computer Science. In addition, he has been awarded 75 U.S. patents. Today, people have made a 180-degree change in Dr. Vasi Philomin and his career, thinking that he is a genius in the field of artificial intelligence

Dr. Vasi Philomin has worked at Amazon Cloud Technologies for more than six years and created a business unit with artificial intelligence and machine learning as its core business. During this time, he participated in the output of many technical services, including language services, visual services, as well as industrial monitoring, enterprise search services, etc.

Think about the concept of opening and closing

Vasi Philomin's opening remarks left a deep impression on people. He discussed issues that everyone was concerned about. When generative AI enters rampant mode, how does Amazon Cloud Technology’s large model compare to other models? In addition, he also mentioned how to ensure a trusting relationship with customers

Dr. Vasi Philomin believes that generative artificial intelligence is not an independent technology, but requires end-to-end integrated technology stack capabilities. This is the fundamental reason why Amazon Cloud Technology builds comprehensive service capabilities, from the bottom layer, the middle platform layer to the upper layer applications, all have been fully considered. For users, there is no need to care about infrastructure and platform tools. They only need to choose according to their own business needs and apply large models directly to enterprise production scenarios

A large model cannot meet all needs, so a large model that truly meets user needs should be a model family platform with multiple products that can support different scene requirements, such as text, images, etc. Even if a model can solve multiple needs such as text and images, users prefer to have multiple choices and choose a large model that suits them based on factors such as actual latency, cost, and accuracy

Take TensorFlow as an example. This is a typical case. TensorFlow is an end-to-end open source machine learning platform. In order to meet the needs of users, Amazon Cloud Technology has optimized it to run on the enterprise's infrastructure. Developers can use TensorFlow on Amazon Cloud Technology to enhance the machine learning (ML) functions of applications and better develop applications in computer vision (CV), natural language processing (NLP), speech conversion, etc.

One problem with using a general large model is that if competitors also use the same large model, the user's technical investment will not be worth the loss. Therefore, Amazon Cloud Technology provides many customized functions based on enterprise private scenarios. For example: AI model text generation model Amazon Titan Text Express and Titan Text Lite, and image generation model Titan Image Generator. Among them, Titan Image Generator can not only generate images based on text, but also has the functions of image editing and hiding watermarks; Amazon Titan Text Express is a "cost-effective" text generation model that allows users to input up to 8,000 word elements; Titan Text Lite is A customized model optimized for copywriting, which can provide input of up to 4,000 tokens

Through the interview with Dr. Vasi Philomin, we can see that what makes Amazon Cloud Technology unique is their correct choice of direction. Amazon Cloud Technology understands that large models should not be the only option, so they choose an open attitude, embrace a wider range of models, and win from an ecological perspective. Behind the broad model, however, there can be difficult issues of control and trust. Therefore, Amazon Cloud Technology has put a lot of effort into ensuring the security, private attributes, and customized attributes of large models. This is not only the advantage of Amazon Cloud Technology in the enterprise field in the past, but also the key to winning customer trust. Therefore, Amazon Cloud Technology has no flaws in the strategic selection of large models. They have an open attitude and a spirit of cooperation, and have established full-stack capabilities

The above is the detailed content of Dr. Vasi Philomin from Amazon Cloud Technology: Seeing the world and the future from the forefront of generative AI. For more information, please follow other related articles on the PHP Chinese website!