Sr SDE, AGI Inference- GenAI

Amazon.com, Inc.•Sunnyvale, CA

39d

About The Position

The Sensory Inference team at AGI is a group of innovative developers working on ground-breaking multi-modal inference solutions that revolutionize how AI systems perceive and interact with the world. We push the limits of inference performance to provide the best possible experience for our users across a wide range of applications and devices. We are looking for talented, passionate, and dedicated Inference Engineers to join our team and build innovative, mission-critical, high-volume production systems that will shape the future of AI. You will have an enormous opportunity to make an impact on the design, architecture, and implementation of novel technologies used every day, potentially by people you know. This role offers the exciting chance to work in a highly technical domain at the boundary between fundamental AI research and production engineering such as Quantization, Speculative Decoding, and Long Context for inference efficiency.

Requirements

5+ years of non-internship professional software development experience
5+ years of programming with at least one software programming language experience
5+ years of leading design or architecture (design patterns, reliability and scaling) of new and existing systems experience
Experience as a mentor, tech lead or leading an engineering team
Experience with inference frameworks such as PyTorch, TensorFlow, ONNXRuntime, TensorRT, LLaMA.cpp, etc.

Nice To Haves

5+ years of full software development life cycle, including coding standards, code reviews, source control management, build processes, testing, and operations experience
Experience with inference frameworks such as PyTorch, TensorFlow, ONNXRuntime, TensorRT, LLaMA.cpp
Proficiency in performance optimization for CPU, GPU, or AI hardware
Proficiency in kernel programming for accelerated hardware using programming models such as (but not limited to) CUDA, OpenMP, OpenCL, Vulkan, and Metal
Experience with latency-sensitive optimizations and real-time inference
Knowledge of model compression techniques (quantization, pruning, distillation, etc.)
Experience with LLM efficiency techniques like speculative decoding and long context

Responsibilities

Develop high-performance inference software for a diverse set of neural models, typically in C/C++
Design, prototype, and evaluate new inference engines and optimization techniques
Participate in deep-dive analysis and profiling of production code
Optimize inference performance across various platforms (on-device, cloud-based CPU, GPU, proprietary ASICs)
Collaborate closely with research scientists to bring next-generation neural models to life
Partner with internal and external hardware teams to maximize platform utilization
Work in an Agile environment to deliver high-quality software against tight schedules
Hold a high bar for technical excellence within the team and across the organization