2024 MIT Digital Technology and Strategy Conference: Lightning Talk - AltaStata

SERGE VILVOVSKY: Hello, everyone. My name is Serge Vivovsky. I'm a former MIT Lincoln Lab researcher, and now CEO of AltaStata. And our mission is securing data collaboration in the AI supply chain. So this is how AI/ML works. Companies in different industries develop applications. They deploy them into data analytic platforms. And at the bedrock, they use cloud storage. So they want to have data-driven decision making in real time and respect data privacy and trust.

It turns out that in 82% of the breaches involved cloud storage data. So this is vulnerable point of this ecosystem, and the data security software market is skyrocketing. According to Google research, if attacker has access to 0.01% of the data set, he can completely destroy the training model. So garbage in, garbage out.

The solution is AltaStata. AltaStata provides a toolkit for human users, devices, and AI/ML analytical platforms that enables to provide data privacy and trust by design. And our product is fortified data lake. Can be implemented in hours in multi-cloud. It provides intuitive user experience, and it can be seamlessly integrated with AI platforms like Databricks, Snowflake, IBM, Watson, AWS SageMaker, and others.

So let's look at the example. The client calls to the customer service, and audio call will be transcribed to the text. And this text will be sent to AI application as a prompt. The application creates the personalized offer, and this offer will be provided to the client in real time. So this data is very sensitive, and LM model is also very sensitive. So it's trained on real financial transaction and real credit history.

So AltaStata can help here to secure audio calls, transcriptions, personalized offers, models, and also the data set for training. So existing encryption solutions fall in two categories. Some provide data privacy by design at cost of integration with AI platforms, and some are opposite. So AltaStata can do both. We provide zero-trust security, and this is applicable for AI/ML.

In addition to that, AltaStata even can boost the speed of the operations because they also compress the data. For example, loading the one gigabyte of encrypted data using AltaStata can be three times faster than just using the same data in the same cloud. And also, data compression can help to lower the cloud storage cost also up to 60%.

So we are MIT-based team. We have US patent for this method, and we've done pilots and we have letter of support by Lockheed Martin and British Telecom. So we are looking for strategic partners. And if you share our vision, we would love to continue this conversation. Thank you.