Magic Leap is an eclectic group of people who share a magical vision of the future. And we’re growing.
Our mission is to harmonize people and technology to create a better, more unified world. Our vision is to amplify the best parts of you and to advance the human spirit.
Job Description
As a Senior Embedded DSP Engineer on the Perception team, you'll be responsible for designing and developing high-performance production software with state-of-the-art computer vision capabilities. Your code will be part of the foundations of our 3D platform that enables the next generation of computing applications for many years to come.
Responsibilities
Develop state-of-the-art performance critical perception systems
Own and drive highly visible performance metrics (e.g. latency, memory, bandwidth, power)
Collaboratively work as part of a fast paced team, analyze and optimize a growing C++ codebase
Help define performance standards for algorithms, develop and drive best practices (performance, maintainability, correctness, etc), lead all stages of the software development cycle and share knowledge
Help drive the integration and development of necessary tooling and automation (profiling, testing/verification, monitoring, etc) that allows the team to move even faster.
Help drive computing platforms for next generation products.
Qualifications
5+ years of professional experience with high performance / low latency applications with modern C++.
Strong CS fundamentals, data structures and algorithms, problem-solving, high quality coding, code optimization, memory optimization, performance analysis, understanding of abstraction and modularity, critical path optimization and tuning.
In depth understanding of computer architecture, computing bottlenecks, memory hierarchy, data coherency, caches, high performance architectures (VLIW, superscalar, etc), SIMD.
Experience with low latency / high bandwidth real-time data pipelines, batch data processing pipelines, low power embedded systems, integration of specialized hardware acceleration blocks.