August 10, 2022
It says something about a company and its people when they drop the process of formulaic job interviews and just let you pitch ideas for the job you want. That’s what happened when I applied to Continual as a Technical Marketing Manager. Five weeks in, I’m pleased to say I’m working on those same ideas, which I’ll detail in a couple minutes.
First, I’d like to describe the singular opportunity we have at Continual. We build a platform for enabling AI at scale across an enterprise. Traditional AI platforms require users to manage complex data infrastructure and write bespoke ML pipelines for each case. We let users build predictive models using a declarative workflow that radically streamlines operational AI at scale. Once you deploy your models, Continual maintains and monitors them so you can move on to your next use case. Your business gets smarter and your life gets easier. That’s a sweet elevator pitch, right? You might like a demo to see how it works. I believe it’s the future of operational AI.
It’s been ten years since I joined Cloudera at the peak of the Big Data age. The size and potential of that market spawned even more new markets and directions, some of which I later saw at Okera, Dataiku, and Confluent. The opportunities around data-driven solutions today are bigger than they’ve ever been, but I’ve also seen the challenges. The potential for realizing more and better value, alas, still comes with intimidating complexity and high failure rates.
In 2013, most people wanted to learn what a data scientist was and what they did. My answer focused on business domain experts who need to acquire these technical skills and have them simplified. They need to build technical teams that help them relate the practices of advanced analytics and data science to business value. And they need the technologies they depend on to meet them more than halfway.
The exciting thing I find in operational AI is how it uses the data to tell the business what rules are in play right now. DeepLearning.AI describes the difference between traditional programming and machine learning like this:
Many of the assumptions of traditional programming were simply constrained by the limits of storage capacity and compute power of their time. Now that we can run any number of compute engines against vast datasets kept in one logical store, the game has changed from coding the business using established rules to learning continually from the models we use to predict future business and behavior.
It’s this change in the way we can apply data that brought me to Continual. Continual brings operational AI to the Modern Data Stack. We empower our customers not just to explore AI, but to deploy it at scale across their business and unlock the full value of their data.
There’s no shortage of choices in the market today. Consider Matt Turck’s “MAD” Landscape, for one, a sprawling who’s-who of vendor offerings across the data processing landscape.
We shouldn’t worry so much about choice, I say. Focus instead on doing your work simply and well. Continual lets you build and deploy continually-improving predictive models — from customer churn to inventory forecasting — without complex engineering and fragile integrations. Define your features and models declaratively. Leverage all the data in your existing cloud data warehouse. Let Continual automate the rest. If you know a better way to operationalize AI at scale, I’d love to hear it.
The Modern Data Stack paradigm highlights similar technologies across the MAD landscape, but it also helps define complementary organizational models. Making AI more accessible and usable by more data practitioners is a strong step in that direction. So is getting your models into production faster, and – one of our big value drivers – continually improving them as the data or your models drift.
Getting solid value quickly from data science projects is tough. Continual is designed to get you there faster, realize value sooner, and let you focus on new use cases as you scale.
I’m here to put our technical messaging in line with our overall marketing message, for one. Also creating a strong UX and reader experience for our documentation – the idea I pitched in my interviews. Working with Bethann Noble, our Head of Marketing, I’ll sum up the rest for now in her words: content, content, content!
We’ll deliver video tutorials over time to make our existing content easier to follow. We’ll build a technical glossary to support the work that Jordan Volz, Brendan McKenna, and others create to support our customers and prospects. It’s an ambitious plan for one person right now, but I’m confident we’ll succeed and grow the team accordingly.
Continual is an exciting place to work. I’m thrilled to be here to help shape the message for operational AI and explore some corners of MLOps I think deserve more light. And of course I’m here to create a “continually improving” user experience for the Continual product and our technical content. If you have questions about Continual and thoughts on how we can improve either one, please reach out.
Andrew Tsao discusses the main reasons he chose to join the Continual team.
Traditional AI platforms require users to manage complex data infrastructure and write bespoke ML pipelines for each case. We let users build predictive models using a declarative workflow that radically streamlines operational AI at scale.