TL;DR: Singularity containers are at DockerCon 2019 this week in San Francisco; stop by to learn about trusted containers for compute from the founder and other members of our dev team, and interact with our cool demo in real time. Experience not required.
Through Thursday of this week, DockerCon 2019 at the Moscone Center in downtown San Francisco remains the focal point for all-things containers. A team from Sylabs has been there all week – exhibiting in the startup area (booth #SU7), talking in the hallways, and more.
If you’ve already dropped by the booth, you may have met Singularity and CentOS Linux founder Gregory Kurtzer, who is Sylabs’ CEO. The leader of the Singularity core development team, Michael Bauer, as well as software engineer Eduardo Arango, are also at the event.
Another reason to drop by: Our demo. Briefly, it’s a badge-detection demo that bridges the analog and digital realms. A camera ‘looks’ at you and your event badge, identifies the text on your badge, and makes a digital representation of it. Under the hood is a containerized implementation of object detection via TensorFlow that allows us to locate a person and their badge, and ultimately recognize characters on the badge via Optical Character Recognition (OCR) using Tesseract OCR. The demo was developed by Sylabs’ software engineer Carl Madison.
Cool in and of itself, the demo truly showcases those use cases for which Singularity is a perfect fit – i.e., those involving compute-driven workloads that need to be executed securely. Because Deep Learning frameworks like TensorFlow make routine use of GPUs for their screaming processing capabilities, the Singularity container Carl built taps into these special purpose devices with the greatest of ease. (Translation: Complex incantations and integrations are NOT required.)
Carl packaged the entire demo, including pre-trained models he built expressly for it, into a single container – a single container that captures the demo through a single Singularity Image Format (SIF) file. Packaged up in this way, Carl made the SIF file that literally encapsulates the demo available via the online Container Library provided by the Singularity Container Services. In so doing, Carl simultaneously made the demo available to the on-site team at DockerCon.
The intersection of AI use cases and container science is an area of significant interest at Sylabs – in fact, we recently joined the Linux Foundation Deep Learning Foundation to assist in broadening and deepening our involvement in this area, and reciprocally ensuring that container science’s impact on AI is well represented. Carl’s demo is one he and other members of the team are continuing to develop – for example, to illustrate use cases in which Singularity containers are orchestrated by Kubernetes or HashiCorp Nomad within the enterprise.
If you’re in the Bay Area this week we hope you’ve made the time to take in DockerCon 2019; if you haven’t yet, there’s still time! From meeting key members of the Sylabs team, to seeing our analog-digital bridging AI demo live, there are lots of reasons to stop by the startups area at Moscone.
Trusted containers for compute. That’s our tagline. If you have use cases that are compute intensive, and require the highest levels of security, you’ll definitely want to check out Singularity containers. We look forward to meeting you at DockerCon!