InfinStor is a multi-cloud Machine Learning platform focused on reducing the failure rate of Machine Learning projects by improving the quality of the models.
Writing credits go to Amit from The Scroll newsletter. Sign up here: thescroll.co
Patent pending technology.
InfinStor’s patent pending storage and compute innovations enable Data Scientists to train using bigger datasets at lower compute costs. The company’s unique storage features such as InfinSnap and InfinSlice ease the management of ML data. InfinStor provides outstanding cost savings for customers by automatically relocating training and inference tasks to the cheapest GPU spot instances across cloud platforms such as AWS, Azure or Google Cloud.
Silicon Valley software engineering veteran Jagane Sundar, a successful serial entrepreneur, has teamed up with former Cloudera exec and Hadoop guru Jitendra Pandey to create InfinStor to unlock the power of Deep Learning.
Company’s tech is in a tailwind.
In a post-pandemic economic recovery, Artificial Intelligence and Machine Learning, which were already growing rapidly in usage prior to the downturn, will accelerate even more. There are voracious new appetites in the market for remote and virtual experiences which will require software that is more efficient and “intelligent.” Yet, the ability to understand and effectively exploit the complexities of machine learning algorithms is an arcane and rare skill.
SDK to the rescue.
InfinStor’s tech allows the user to leverage ML and AI without the steep learning curve right out-of-the-box.
InfinStor co-founder and CEO Jagane Sundar explains, “to cite one example, when it comes to operating in the cloud, all clouds at all times are not equal. Pricing varies greatly depending on a wide range of variables — load, time of day, etc. Machine Learning activities should be performed in spot instances as much as possible to take advantage of pricing variance, and pricing between the clouds — AWS, MS Azure, etc. — should be considered carefully before each training run. But today’s ML tools do not provide the capability to run workloads in spot instances, or to move workloads between clouds.”
The bottom line. By using InfinStor in an AI/ML application, hosted cloud spot pricing and cloud-to-cloud transfer can be exploited, greatly reducing operating costs.