DeepScale was founded by the deep learning researchers from UC Berkeley who created SqueezeNet. DeepScale is developing perception systems that enable automated vehicles to interpret their environment in real-time using low-cost hardware.
What you will be doing
In this position, tasks that you will be assigned to at DeepScale include:
- Developing fast implementations of deep learning kernels on emerging accelerator architectures
- Designing and implementing compiler transformations for our in-house neural network representation
- Working with model developers to balance performance and accuracy
What you must bring to the table
- A PhD in electrical engineering, computer engineering, or computer science. Or, 10+ years of industry experience.
- A track record of writing really fast code. (This can be for embedded platforms, accelerator platforms, or supercomputers.)
- Published papers (e.g. in Supercomputing, IPDPS, or PPOPP) and/or open-source code that demonstrate your skills in writing fast code.
Nice to have
- Original research contributions in the field of programming languages and compilers
- Original research contributions on the topic of developing faster or more scalable algorithms
What DeepScale brings to the table
- Career Growth: An opportunity to learn how to build an impactful product while continuing to advance your research career.
- People: A team of 20+ engineers (7 of whom have PhDs as of this writing) to collaborate with.
- Get hands-on experience in entrepreneurship and commercialization of Deep Learning technologies.