Orchestrated Deploy of StarCluster
This archive was generated by
I’m a contributor to Apache Mesos and work for a company called Mesosphere that creates Mesos-oriented products. For anyone that doesn’t know, Mesos is a resource orchestrator for large cluster deployments. It manages compute nodes and distributed computations.
As a hobby project i’m implementing a variety of deep learning systems onto Mesos. The idea is that as the tools for resource orchestration increase and the barrier to entry for deep learning decreases, a large influx of ‘casual’ deep learning environments will enter the ecosystem. Furthermore, i’m trying to show measurable speed increases to DL systems through new distributed architectures.
However, before I do any work on this, I was curious as to whether this is a relevant project to anyone here, and what the star cluster deployments at MIT look like (obviously don’t need to go into research specifics here, just maybe size and capacity).
Project Manager, Growth Team
BS in Electrical Engineering (2012)
Received on Mon Jun 29 2015 - 17:21:20 EDT