A vision to power games and life

The last thing data science and Artificial Intelligence (AI) experts want to worry about is infrastructure. Thankfully , Tridib Mukherjee, the Chief Data Science & AI Officer at Games24x7, doesn't have to. “Partners like Amazon Web Services (AWS) help us focus on solving business problems and creating great user experiences without worrying about computing infrastructure,” he said.

AI: then and now

This partnership is a strategic win for both parties, especially for Tridib, who is dealing with very large scale datasets. Everyday millions of transactions pour in, generated by millions of gamers on Games24x7 - a leading tech company with a portfolio that spans online skill games.

Tridib’s career path is one guided by passion, perseverance and sportsmanship. As a sports enthusiast, he was naturally drawn to gaming. The opportunity to solve problems at scale and convert raw data into actionable insights was another big draw. “The first algorithms aimed to create the best chess player or master the Chinese board game Go. AI has evolved significantly since those early days,” said Tridib, from his Bangalore office.

Around the time experts were developing systems to outsmart humans in games like chess, Tridib was diving into AI during his computer science studies at Jadavpur University. “There was a course on AI. Back then, and now, the core objective has remained the same: to intelligently find something that human intelligence can find, via machines or computing agents. That was my first exposure to AI,” he recalled.

Graduating in 2000, Tridib pursued a PhD at Arizona State University. One of his early projects involved generating thermal maps of server rooms using sensor data. He proposed neural networks for this task, but the compute resources in 2005-06 couldn't handle the workloads. “It was impractical then, but now, with better resources, the power of Large Language Models (LLMs) and advanced AI capabilities, it’s a different story. Yet, the goal of mimicking human intelligence remains unchanged,” said Tridib.

Enhancing gaming with AI

After a stint in academia, Tridib joined Xerox Research Labs in the US and India, before moving to Games24x7 in 2018. “Games24x7 was my first consumer-facing product company, offering a significant opportunity in AI. We started with data science, then expanded to data-driven decision-making and machine-driven algorithms for hyper-personalization and similar tasks,” he shared.

At Games24x7, the focus is three dimensional — on hyper-personalization, gamification and responsible gaming. With millions of users, understanding player intent from generated data is crucial. It's a complex challenge, like looking for a needle in a haystack. Responsible gaming is about identifying people (less than 0.5% of the users) who might be anomalous, but is essential for a safe gaming experience.

The company also leverages AI to help in fraud and bot detection by identifying any anomalies or spotting unusual activities. This ensures that a human player is not pitted against a bot, but is playing against a fellow human.

Beyond ensuring responsible gaming at the company, AI significantly optimizes workflows, aiding in business decision-making, such as helping marketing teams manage budgets and spending. Additionally, intelligence and automation in different workflows have led to productivity improvements, like more effective responses to customer queries. “My charter is to drive the entire AI functioning for the organization. Typically, the success rate for AI projects, when they move from Proof of Concept (PoC) to production is 15-20%. Our success rate is 80-85%, and that’s what I drive,” said Tridib.

This impressive success rate, well above the industry average, wouldn't be possible without the right infrastructure and technology tools. A key contributor is AWS SageMaker, a cloud-based Machine Learning (ML) platform that facilitates the creation, training, and deployment of ML models. SageMaker can also deploy ML models on embedded systems and edge devices.

Tridib adopted SageMaker, early on, to build Games24x7’s in-house AI modeling capabilities. “SageMaker allowed us to seamlessly train and test models. It provided the framework, so we didn't have to create the infrastructure or APIs ourselves,” he explained. All data resides on Amazon S3 (Amazon Simple Storage Service), a highly scalable, durable, and secure object storage service designed for industry-leading scalability, data availability, security, and performance.

“One of the first things we identified at inception was the need for proper technology infrastructure. Of course, this infrastructure continuously evolves in terms of capabilities. On this front our journey started with AWS SageMaker,” said Tridib.

Another service Tridib spoke of is Bedrock, a fully-managed service that offers a choice of high-performing Foundation Models (FMs) from leading AI companies like Anthropic, AI21 Labs, Meta, Stability AI and Amazon, through a single API, along with a broad set of capabilities that can be used to build generative AI applications. With Amazon Bedrock, users can experiment with and evaluate top FMs for their use case, customize them with their data and build agents that execute tasks using their enterprise systems and data sources.

A natural evolution

Even as AWS keeps Tridib up to speed with the latest trends and capabilities, he sees AI evolving naturally, much like any research field. Previously, companies solved their problems through data and a library of algorithms that trained the model to find a solution. The major shift in recent years is the concept of pre-trained models.

“This concept has transformed the way one looks at AI in general. Because now you are talking about a library of AI models and the exponential increase in those kinds of models with new capabilities. This has democratized models, and the application of models to solve various problems. This will lead to custom models, pre-built for certain industries,” he said, regarding the future of AI. He also anticipates a lot of standardization in terms of how the AI platform should work; how different models can perform in different ways, vulnerabilities in AI systems and the creation of guardrails in the system. Tridib expects a lot of movement in this direction, including regulation, that will eventually reduce the failure rates of AI models.

When he isn’t revolutionizing the gaming industry with AI, Tridib likes to watch football, cricket or a movie in his free time.

Keeping apace with AI

There's a lot of excitement among users to make AI work for them. Alongside improvement in AI outcomes, AI literacy will also improve. That is, people will be a lot more aware of how to use AI and what to expect. While Tridib works with AI and stays ahead of the curve with new developments, there are quite a few Amazon services which come in handy to help him keep abreast with the latest news in this rapidly evolving space.

These services include Amazon Q Developer. It is a generative AI-powered conversational assistant that can help users understand, build, extend, and operate AWS applications. Users can ask questions about AWS architecture, AWS resources, best practices, documentation, support, and more. Amazon Q is constantly updating its capabilities so the questions get the most contextually relevant and actionable answers.

“Amazon Q developer is definitely going to make a dent in the way development happens from an engineering point of view and the way code reviews happen. A lot of manual effort will get reduced,” said Tridib,

While human-machine interactions are increasing with AI, Tridib doesn't see humans being replaced by AI. However, people who use AI will definitely have an edge over non-users. On the future of AI and human interactions Tridib noted, “Humans will be interacting with AI to fast-track their work. The large language models have been able to create fantastic systems to understand language dynamics. But the analytical capability, emotional thinking that humans have is yet to be achieved by machines. A lot of research is going on in this space. Until then, AI will continue to be a great assistant to humans.”