As we prepare for ScyllaDB Summit 2019, we are producing a series of blogs highlighting this year’s featured presenters. And a reminder, if you’re not yet registered for ScyllaDB Summit, please take the time to register now!
Today we are speaking with Timo Mechler, Product Manager, and Charles Adetiloye, Machine Learning Platform Engineer, both of SmartDeployAI, who will be co-presenting the session Simplifying the Creation of ML Workflow Pipelines for IoT Application on Kubernetes with ScyllaDB.
Machine Learning, workflow pipelines, IoT, Kubernetes… That’s quite a buzzword compliant title! You sure you can fit all that into a single ScyllaDB Summit session?
Timo: Well, we sure hope so! 😃 While the title is buzzword heavy, the technologies we are talking about are at the top of everybody’s mind today in the Enterprise IT space today. We wanted to show the audience how we have used these technologies together to build a novel software platform that we hope will continue to democratize the field of artificial intelligence (AI).
Charles: Yeap, it sure sounds like lots of buzzwords, because a lot of people are still trying to figure out all these new things, the technology frameworks are evolving rapidly in different directions that it’s quite difficult to establish the ground-truth most of the time! We have seen this over and over from all various client engagements. But the value proposition we bring to the table is a clear understanding of all these technology components (not just a buzzword to us 😊) from architecture to implementation, integration, and deployment at scale.
One more thing if I may add is that we don’t just jump on these new technologies — for instance, Kubernetes, ScyllaDB, etc., as they show up — we do a comparative analysis with other parallel technologies. We check the long term viability and most importantly evaluate how well we can use it to build a better product to serve our customers. It took a little while for us to come around to ScyllaDB but we are loving it right now!
Help our audience understand the nature of your talk. What will you be covering in specific? Who would benefit most by attending your session?
Timo: We will be talking about the creation of IoT data pipelines for artificial intelligence and machine learning and how we are running these on Kubernetes with the help of ScyllaDB to drive efficiency and performance at scale.
Charles: Our talk will be centered around operationalizing pipelines for IoT systems with AI capabilities for real-time inferencing. It’s very easy to fall into the trap of building a Monster-AI pipeline system, that is not easily scalable, not cloneable; i.e you can’t instantiate a new pipeline on the fly, or requires database tuning every other day! We will be talking about how we have solved all these problems by leveraging Kubernetes and ScyllaDB for the efficient utilization of computing resources and storage.
Tell us more about how SmartDeployAI evolved. When and how did it get its start?
Timo: SmartDeployAI grew out of our own experience consulting in the big data space by building data pipelines and analytical models (e.g. AI and ML) for clients. We noticed that these clients often lacked the necessary infrastructure or DevOps know-how to be able to put these types of models and pipelines into production and then run them at scale in their enterprise. This served as our “lightbulb moment” and decided to focus our energy into building a collaborative and intuitive software solution that would make it easy for anyone to run AI & ML at scale, without the need for specialized (and hard to find) DevOps skills, or a multi-million dollar budget.
Charles: Yeap, Timo said it well, This is a result of all our years of consulting engagement in this space. I like to use this analogy: The Journey from point A to B could be a straight line dash, but it’s very disturbing when you see lots of companies expend resources and time but they don’t make it to the destination. Our goal is to help our clients eliminate these distractions and get them to the destination with minimal effort and at a fraction of the cost.
What do you believe is the most compelling technical aspect of the architecture?
Charles: Its simplicity and intuitiveness. We treat components of the pipeline we build like a “Lego” block with the right level of functional abstractions.
Where can people learn more about SmartDeployAI?
Timo: A good place to start would be our website – www.smartdeploy.ai. We are also active on Social Media and have our own Slack Channel (http://bit.ly/ai-pipelines) for those that are interested in learning more about our platform and/or collaborating with us on building more data pipelines for AI & ML.
Outside of your day-to-day jobs, what interests or hobbies do each of you have?
Timo: Well, being a father or two young children does keep me busy, but I do like to step away from the keyboard as much as I can and enjoy a good CrossFit workout. I used to work in finance and still like to keep my pulse on the markets as time allows. Other than that, I’m a big fan of classical music – I enjoy visiting live performance, and I’m continuing to grow my own personal collection of recordings.
Charles: Being a dad keeps me busy and spending time with family is what I do most of the time when I am away from the keyboard. Other than that, I love things that can fly, Planes and drones! Working on AI and MLOps has taken a lot of my time in the last few years but I need to go back and get my VFR license so that I can get back flying my favorite plane – the Cessna 172.
Thank you both for taking the time to talk to me today. We’re looking forward to your ScyllaDB Summit talk!