Inside this Business Group
What you will be working on
• Develop tools and models for our custom 16 bit numerical format for deep learning
• Invent optimized architectures and layers for training convolutional and recurrent networks in low bit-width using neon, our Python Deep Learning framework
• Work on applications in the following areas: computer vision, object detection and localization, speech recognition, natural language processing, recommendation systems, forecasting, and multimodal learning
• Instrumenting our library to detect and track overflows, monitor dynamic range
Who we are looking for
• Candidate must be pursuing a Bachelor's or Master's in computer science, electrical engineering or related fields (statistics, applied math, computational neuroscience).
• Minimum of 1 year experience in the following areas:
Deep learning, machine learning & big data. Prior experience with binary networks or float16 is a plus.
• Strong skills using fixed point or low precision numerical formats for algorithms in fields such as image and audio processing. Ideally this includes experience porting floating point algorithms to fixed point
• Experience with SIMD acceleration using frameworks such as AVX, OpenCL, CUDA is a plus.
• Experience designing professional software using Python, C++