Software Development Engineer, Neuron Collectives, Annapurna Labs
Core
Optimize collective operations and communication patterns for AWS Trainium AI training chips to scale compute across data center topologies.
Role type
Senior IC software engineer (systems/hardware integration)
Builds
High-performance collective communication software for AI training clusters
Domain
AI infrastructure / High-performance computing / Hardware-software integration
Deliverable
production ML models
Required skills
C/C++, distributed systems, collective algorithms, hardware/software integration, performance optimization, DMA/firmware interfacing
Preferred skills
real-time systems, all-reduce/all-gather algorithms, distributed training frameworks
Technologies
C/C++, Neuron Explorer, Trainium
Responsibilities
Enhance collective algorithms and topologies for optimal training performance; Use tools like Neuron Explorer to identify bottlenecks; Monitor and analyze processor, DMA, firmware, and workload metrics; Optimize collective operations to scale AI compute; Work closely with hardware team to co-optimize software and silicon; Develop and optimize C/C++ implementations of collective communication patterns; Build and maintain analysis frameworks and automation solutions