Hello World (1)
We have been working on something exciting and we want to introduce our work through a series of blogs. But first the context…
Machine Learning and Artificial Intelligence promise enormous value creation across industries and use cases, but achieving that promise will require reducing entry barriers and democratizing utilization. Today, deploying ML/AI techniques to solve problems at scale¹ requires employing a highly-skilled community, consuming millions of dollars annually. In this structure, only large and/or well-resourced companies stand to benefit. We aim to lower the entry barriers for firms and individuals, providing a platform that both lowers the required skill level to work with ML/AI and greater leverages the skills of highly trained personnel.
We further recognize that most businesses and individuals have business domain expertise in their respective field, have access to data, and know what problems they need to solve. But, using ML/AI to solve problems is difficult and is not easily scale-able. To do so, companies must build a team of trained Data Scientists that have a unique combination of math, coding skills and the business domain expertise to solve high value problems. This combination of skill sets is extremely rare, and those capable are well-compensated. With the demand for ML-based solutions outpacing supply, teams are growing exponentially, further stretching supply.
With the right team in place, companies then face the daunting challenge of managing an extremely complex ML life-cycle, from model development to solution deployment. Today, this cycle is often fragmented, silo-ed, iterative, inefficient, and time consuming. For example, data science teams often spend as much as 80% of their time simply cleaning the data — getting the “data” to do the “science”.
To tackle this very problem, we have embarked on mission to democratize Machine Learning and Artificial Intelligence.
To achieve our goal, we have been hard at work building an AutoML platform (Augmented ML to be really specific as one cannot take human out of the loop) that takes advantage of Bayesian Optimization, Evolutionary Strategies and best of class visualizations to make it easier for the Business user to build, interpret and deploy predictive models without writing code.
Our approach to building the platform lies on the principal that the fundamental Machine Learning workflow is iterative creating opportunities to automate and simply. The prototypical ML workflow looks something like this:
The Platform does this:
We want the Business Analyst or the Business Intelligence user to have access to the power of ML to enable predictive analytics without relying on the already stressed Data Science groups in their organizations. We want Data Science groups to accelerate their work through speed and efficiency.
In the next blog, we will introduce the Platform and its capabilities in details.
Going Live — We are now in Beta and one can sign in to become a beta user at our website.
These are exciting times for us and hopefully we can make it as exciting to the vast community out there.
Let me end the post by a quote that I feel is relevant to our mission.
“Not everyone can become a great artist, but a great artist can come from anywhere.” — Anton Ego
[1]: https://www.gartner.com/smarterwithgartner/gartner-top-strategic-predictions-for-2019-and-beyond/
