The interdisciplinary field of data science is an incredibly complex and sometimes nebulous industry. It was recently described by NVIDIA CEO Jensen Huang as “the new fourth pillar of the scientific method.”[1] As exciting as that is, it’s long been a buzzword that gets used too much, too loosely, and has no single agreed-upon definition. In fact, the field is so broadly defined that even graduate programs for data science have different curriculum requirements.[2]

To define it broadly, data science is a series of computational methods used to extract knowledge and insights from large amounts of data. For most people, it’s that predictive quality that makes it so exciting, not to mention useful. Analyzing petabytes of data won’t do you any good if it won’t help you decide what to do next.

In this new era of science fiction becoming science fact,[3] how to maximize your specific workflow as a data scientist isn’t always obvious. More and more companies are finding themselves with enormous amounts of data, and they require more professionals armed with new techniques to make sense of it. However, figuring out the right hardware solution should be the easy part.

Your first foray into the intimidating world of data analytics, recurrent neural networks, or machine learning algorithms doesn’t need to cost six figures. Not everyone is going to need an NVIDIA DGX-1 right off the bat. Basically, you need a stable product with lots of CPU cores and plenty of room for GPUs, since you’ll also need to be able to keep up with demanding GPU-accelerated tasks and intensive AI-related applications like TensorFlow, Caffe2, or PyTorch.

In a field that is easily perplexing, as a professional you may also like to have a little extra reassurance. After all, there’s no guarantee something off the shelf will be optimized for your workflow. Figuring all that out on your own takes time that could be spent elsewhere. It would be great to find a system pre-installed with machine learning libraries on top of software that’s already been optimized by NVIDIA themselves.

There isn’t an industry out there not being affected by innovations in AI, and you need a quiet, compact machine specifically designed with the modern data scientist in mind to reduce time to insight and maximize your productivity. Unfortunately, such a thing has only existed in a data scientist’s dreams. Well it’s time to wake up, because BOXX engineers (working closely with NVIDIA) have designed such a system, and are proud to be offering the data scientist’s dream machine, APEXX W3. Whether you’re designing self-driving cars, analyzing genomes, or just want to make a better killer robot,[4] it’s the ideal candidate for desk-side deep learning. Equipped with an 18-core Intel® Xeon® W processor (up to 36 threads), 512GB of memory, and as many as four, full-sized NVIDIA Quadro® GTX™ GPUs, this beefy machine is purpose-built for deep learning development.

For the more experienced data scientist, or if 18 cores just aren’t enough, the APEXX D4 allows for up to 56 cores across two CPUs, 2TB of memory, and additional room for several high-capacity drives.[5] (It’s perfect for complex rendering and broadcast graphics, too, if that’s more your thing.) Add a Quadro RTX 8000, coupled with NVIDIA’s brand new CUDA-X libraries for accelerating AI applications, and you’re looking at a beast of a machine capable of solving the future’s problems before they even happen.
[1] GTC 2019.
[2] Seriously.
[3] Did you know the Holodeck is a real thing now?
[4] BOXX Technologies does not advocate the use of killer robots.
[5] Check out the spec sheet for more details.