Frequently asked questions.
Find the answer to one of the questions below:
We help our clients to optimize their computational tasks by moving them from a higher-order programming language to an efficient WebAssembly implementation. Next, our platform makes sure that the task is available at any time, at any scale, and at the lowest cost possible.
For example, your data science team has created a machine learning model to predict the churn probability of your customers. We will work with your data science team to optimize this machine learning model, such that it is as fast and efficient as possible. Next, we will host the task on our platform such that you can use it in production: you can now readily receive churn-rate predictions for all your customers any time you want without having to worry about the underlying technology. We will make sure that we scale with your demand, and that we provide your computations as fast and cheaply as possible.
For us, a computational task is any operation that is done on a bit of input data and outputs some result. Now, that's cryptic, so, let's look at some examples:
Effectively, anything that can be carried out in a serverless computing framework can be considered a computational task. This includes pretty much all supervised learning models and thus includes the fast majority of contemporary Artificial Intelligence (AI) applications.
Whether you are already a frequent user of computational tasks (because, for example, you use several machine learning models to power your online services) or just getting started, optimizing your tasks is useful for several reasons:
We use WebAssembly (WASM when run in a browser, WASI when run on a server) to optimize your task. Why? Well, because
Once you have an optimized scailable-task, you can effectively run it anywhere you want. However, our platform allows you to put your task into production effortlessly. Our platform will scale with your demands, and it will use Machine Learning methods to make sure that your task is carried out as efficiently and cheaply as possible (obviously within your privacy and security requirements). With our platform, you do not have to worry about mainteinance of your tasks, nor do you have to worry about vendor lock-in with one specific (cloud-) compute supplier (such as AWS or Google). We will maintain your task for you, and we will get it executed at the best possible compute-node.
Yes, off course we can! Our team can work closely with your data science or machine learning team after they have developed their first proof-of-concept, or we can help you from scratch. We have decades of machine learning experience behind us, and we will be able to help you identify useful tasks and create them.
With most of our clients we start when the Data Science team (or AI or ML team) has their data available to them and has some idea how they would like to use modern AI and ML methods. This can be for simple tasks such as compressing images on the fly, or much harder tasks such as high-dimensional prediction problems or clustering problems. Our team then joins up with yours to jointly finalize the task, and create an highly efficient version of the task. Next, we put your task in production on our platform, and you can use it anytime you want, at any scale you want.
Many of our tasks are effectively created through machine learning; we put trained machine learning models in production. Now, in our view, a lot of currently applied AI is simply supervised learning, albeit with a preference for deep neural networks. Hence, a lot of our computational tasks are actually AI.
Well, no. We have to start and stop somewhere. We will work with you to create tasks, and we will put them into production on our platform. However, if you are new to the AI / ML / Data Science world and you still have to structure the access to your (internal) data, that is not our expertise. We obviously do have friends that can help you with this. If you are interested, please send us a message.
We make sure that each scailable-task is stored in our secure vault. When you consume your compute task the privacy and security are ensured in several ways:
https, but we can provide additional encryption if necessary.
Note that the technology we use for our webnodes is inherently safe; WebAssembly is fully sandboxed and hence whatever task we might distribute to a users machine cannot affect the state of that machine. Thus, our suppliers of compute are safe as well.