AI Systems Engineer – AI Model (Training & Inference)

Advanced Micro Devices, Inc•Markham, ON

1d•Hybrid

About The Position

ADVANCE YOUR CAREER. ADVANCE THE WORLD. At AMD, we believe technology can change lives for the better. It can heal us, entertain us, and make us more connected, productive, and understanding of the world around us. And we’re looking for talent who feel the same: people who want to leave the planet better than they found it, those who don’t shy away from humanity’s challenges but are determined to help solve them. AMD is powering the next generation of supercomputing, high-performance computing, cloud, and AI. Whether you’re designing next-gen processors, enabling AI breakthroughs, or creating go-to-market plans, every role at AMD contributes to something bigger — technology that moves the world forward. THE ROLE: The AMD AI Group is looking for a Senior Software Development Engineer to own the end-to-end model execution stack on AMD Instinct GPUs - spanning training infrastructure at scale and high-performance inference serving. THE PERSON: This role demands someone who has shipped LLMs on real hardware, written GPU kernels that moved production metrics, and built the systems infrastructure (orchestration, storage, monitoring) that keeps thousands of GPUs productive. You will be instrumental in ensuring AMD GPUs are first-class citizens for frontier model training and inference across current and next-generation Instinct accelerators.

Requirements

Shipped LLMs on real hardware
Written GPU kernels that moved production metrics
Built the systems infrastructure (orchestration, storage, monitoring) that keeps thousands of GPUs productive
Industry experience shipping production AI/ML infrastructure, with hands-on work spanning both training and inference.
Direct experience enabling frontier models (GPT-4 class) on AMD Instinct hardware end-to-end.
Familiarity with AMD Instinct MI-series architectures (MI300X, MI350X, MI355X) and RCCL communication library.
Experience designing validation frameworks, proxy benchmarks, or synthetic workload suites for GPU infrastructure at scale.
Bachelor’s or Master’s degree or Ph.D in Computer/Software Engineering, Computer Science, or related technical discipline

Nice To Haves

Background in building anomaly detection, log analysis, or observability systems for large-scale distributed GPU infrastructure.
Contributions to open-source AI frameworks (PyTorch, vLLM, SGLang, DeepSpeed, Megatron-LM).
Experience with pre-silicon software validation or hardware-software co-verification workflows.
Publications or patents in HPC, ML systems, or GPU kernel optimization.

Responsibilities

Enable and optimize large-scale model training (LLMs, VLMs, MoE architectures) on AMD Instinct GPU clusters, ensuring correctness, reproducibility, and competitive throughput.
Build and maintain training infrastructure: job orchestration, distributed checkpointing, data loading pipelines, and storage optimization for multi-thousand GPU clusters on Kubernetes.
Debug and resolve training-specific issues including gradient norm explosions, non-deterministic behavior across GPU generations, and compute-communication overlap in distributed training (FSDP, DeepSpeed, Megatron-LM).
Optimize RCCL collective communication patterns for training workloads, including all-reduce, all-gather, and reduce-scatter across multi-node topologies.
Develop monitoring, alerting, and compliance infrastructure to ensure training cluster health, data security, and SLA adherence at scale.
Design and build end-to-end validation and testing infrastructure using proxy workloads, synthetic benchmarks, and configurable workload generators to systematically validate platform readiness across AMD Instinct GPU generations.
Write and optimize high-performance GPU kernels (GEMM, attention, quantized matmul, GPTQ/AWQ) in HIP, Triton, and MLIR targeting AMD Instinct architectures, with demonstrated ability to outperform open-source baselines.
Drive end-to-end inference enablement on new AMD GPU silicon - be among the first to get frontier models running on each new Instinct generation, creating reproducible guides and reference implementations.
Optimize inference serving frameworks (vLLM, SGLang, TorchServe) for AMD GPUs: batching strategies, KV-cache management, speculative decoding, and continuous batching for production throughput/latency targets.
Develop novel approaches to inference acceleration, including bio-inspired algorithms, SLM-assisted batching, and custom scheduling strategies that exploit AMD hardware characteristics.
Build quantization pipelines (FP8, FP6, FP4, GPTQ, AWQ) for production model deployment, ensuring quality-performance tradeoffs are well-characterized across AMD GPU generations.
Collaborate with AMD silicon architecture and pre-silicon teams to provide software feedback and validate software stack integration on next-generation Instinct GPU designs for both training and inference workloads.
Build observability and automated analysis tooling: log analysis pipelines, anomaly detection, performance baselining, regression detection, and diagnostic workflows for large-scale GPU clusters.
Contribute to the open ROCm ecosystem and AMD's developer experience — SDKs, CI dashboards, documentation, and developer cloud enablement.