← Back to jobsJob ID col-cmwpdycob0bevd8ha0lf586eu

AMDPosted 1 day ago

Research Scientist (Generative Modeling)New

Full-time
On-site
$150K – $215K/yr
Posted 1 day ago

AMD is hiring for a role focused on improving the performance of generative models in training and production serving. The work includes GPU kernel optimization, inference and training performance, profiling, and numerical correctness in collaboration with researchers.

Posted
1 day ago
Source
careers.amd.com
Sector
Information Technology & AI

Opens the employer's application page in a new tab

Responsibilities

  • Optimize inference and serving end to end — latency, throughput, batching, caching, and scheduling — to serve our models efficiently at production scale
  • Write and tune GPU kernels (CUDA, Triton) for hot paths; drive kernel fusion, memory- and bandwidth-bound optimization, and low-precision (FP8/INT8) execution
  • Optimize training throughput and GPU utilization: parallelism strategies, communication/compute overlap, mixed precision, and eliminating pipeline stalls
  • Build performance models, profiling workflows, and observability that make throughput, latency, cost, utilization, and their tradeoffs legible across the stack
  • Own numerical correctness across precision, kernel, and hardware changes — treating correctness as part of performance, not separate from it
  • Partner with researchers to productionize models for serving and to make experiments run faster and more reliably
  • Where needed, work on the distributed systems that training and inference run on — but the core of the job is squeezing the most out of every GPU

Qualifications

  • Strong performance-engineering foundations: profiling, roofline analysis, latency/throughput optimization, and disciplined root-cause investigation
  • Deep GPU programming and optimization experience (CUDA and/or Triton) — kernel-level tuning, memory hierarchy, and bandwidth optimization at scale
  • Hands-on experience optimizing inference and serving for large models: batching, KV/prompt caching, quantization, and low-latency, high-throughput sampling
  • Hands-on experience optimizing training performance: parallelism, distributed communication, mixed/low precision, and utilization
  • Working knowledge of ML framework internals (PyTorch and/or JAX; torch.compile, XLA, or similar compiler paths)
  • Strong proficiency in Python, with the ability to drop into C++/CUDA (and Rust or Go) as the work demands
  • High-ownership mindset — you measure yourself by throughput shipped and latency cut, not tickets closed

Preferred

  • Experience at an AI lab or ML-native company, optimizing systems used directly by researchers and productionizing research code
  • Low-precision and numerics depth: FP8/INT8 quantization, mixed-precision, and detecting numerical regressions across hardware platforms
  • Distributed systems for large-scale training and inference — collective communication (NCCL), interconnects (NVLink), model and tensor parallelism, and fault tolerance. A strong plus, but not a substitute for the core skills above
  • Experience serving generative, diffusion, video, or 3D/spatial models — not just text LLMs
  • Multi-accelerator experience (GPU plus TPU or Trainium) and partnering with hardware vendors on accelerator capabilities
  • Building performance-modeling and observability frameworks for GPU utilization and cost
  • AI Lab Experience
  • ML-Native Company Experience
  • GPU Experience
  • TPU Experience
  • Trainium Experience

Required skills

C++CUDADistributed SystemsGPU Programming and OptimizationJAXNCCLNVLinkPyTorchTraining Performance OptimizationTriton

Full job description

AMD is hiring for a role focused on improving the performance of generative models in training and production serving. The work includes GPU kernel optimization, inference and training performance, profiling, and numerical correctness in collaboration with researchers.

Responsibilities

  • Optimize inference and serving end to end — latency, throughput, batching, caching, and scheduling — to serve our models efficiently at production scale
  • Write and tune GPU kernels (CUDA, Triton) for hot paths; drive kernel fusion, memory- and bandwidth-bound optimization, and low-precision (FP8/INT8) execution
  • Optimize training throughput and GPU utilization: parallelism strategies, communication/compute overlap, mixed precision, and eliminating pipeline stalls
  • Build performance models, profiling workflows, and observability that make throughput, latency, cost, utilization, and their tradeoffs legible across the stack
  • Own numerical correctness across precision, kernel, and hardware changes — treating correctness as part of performance, not separate from it
  • Partner with researchers to productionize models for serving and to make experiments run faster and more reliably
  • Where needed, work on the distributed systems that training and inference run on — but the core of the job is squeezing the most out of every GPU

Preferred

  • Experience at an AI lab or ML-native company, optimizing systems used directly by researchers and productionizing research code
  • Low-precision and numerics depth: FP8/INT8 quantization, mixed-precision, and detecting numerical regressions across hardware platforms
  • Distributed systems for large-scale training and inference — collective communication (NCCL), interconnects (NVLink), model and tensor parallelism, and fault tolerance. A strong plus, but not a substitute for the core skills above
  • Experience serving generative, diffusion, video, or 3D/spatial models — not just text LLMs
  • Multi-accelerator experience (GPU plus TPU or Trainium) and partnering with hardware vendors on accelerator capabilities
  • Building performance-modeling and observability frameworks for GPU utilization and cost
  • AI Lab Experience
  • ML-Native Company Experience
  • Research Systems Optimization
  • Research Code Productionization
  • FP8 Quantization
  • INT8 Quantization
  • Mixed-Precision Computing
  • Numerical Regression Detection
  • Cross-Platform Numerical Analysis
  • Distributed Systems
  • Large-Scale Training
  • Large-Scale Inference
  • NCCL
  • NVLink
  • Model Parallelism
  • Tensor Parallelism
  • Fault Tolerance
  • Generative Model Serving
  • Diffusion Model Serving
  • Video Model Serving
  • 3D Model Serving
  • Spatial Model Serving
  • GPU Experience
  • TPU Experience
  • Trainium Experience
  • Accelerator Capabilities
  • Performance Modeling
  • GPU Utilization Observability
  • Cost Observability
  • Hardware Vendor Collaboration

Company

  • High-performance computing solutions for data centers, AI, PCs, gaming
  • Santa Clara, California, USA
  • http://www.amd.com/
  • Founded in 1969
  • 10001+ employees

Qualifications

  • Strong performance-engineering foundations: profiling, roofline analysis, latency/throughput optimization, and disciplined root-cause investigation
  • Deep GPU programming and optimization experience (CUDA and/or Triton) — kernel-level tuning, memory hierarchy, and bandwidth optimization at scale
  • Hands-on experience optimizing inference and serving for large models: batching, KV/prompt caching, quantization, and low-latency, high-throughput sampling
  • Hands-on experience optimizing training performance: parallelism, distributed communication, mixed/low precision, and utilization
  • Working knowledge of ML framework internals (PyTorch and/or JAX; torch.compile, XLA, or similar compiler paths)
  • Strong proficiency in Python, with the ability to drop into C++/CUDA (and Rust or Go) as the work demands
  • High-ownership mindset — you measure yourself by throughput shipped and latency cut, not tickets closed
  • Experience at an AI lab or ML-native company, optimizing systems used directly by researchers and productionizing research code
  • Low-precision and numerics depth: FP8/INT8 quantization, mixed-precision, and detecting numerical regressions across hardware platforms
  • Distributed systems for large-scale training and inference — collective communication (NCCL), interconnects (NVLink), model and tensor parallelism, and fault tolerance. A strong plus, but not a substitute for the core skills above
  • Experience serving generative, diffusion, video, or 3D/spatial models — not just text LLMs
  • Multi-accelerator experience (GPU plus TPU or Trainium) and partnering with hardware vendors on accelerator capabilities
  • Building performance-modeling and observability frameworks for GPU utilization and cost

Required skills

  • C++, CUDA, Distributed Systems, GPU Programming and Optimization, JAX, NCCL, NVLink, PyTorch, Training Performance Optimization, Triton

Job details

Employment type

Full-time

Seniority level

Not stated

Date posted

Nov 20, 2026

Work mode

On-site

Job source

careers.amd.com

Company

AMD

Information Technology & AI

High-performance computing solutions for data centers, AI, PCs, gaming

See all 4 jobs from AMD →
  • Santa Clara, California, USA
  • Information Technology & AI
  • 10001+ employees
  • Founded 1969
  • amd.com

Similar jobs

Matched on role, skills and location · live jobs only

See more Machine Learning, Model Training and Inference jobs in San Francisco

Bright Vision Technologies

1d ago

New

MLOps Engineer

Full-timeRemoteUnited States$115K – $165K/yr

Bright Vision Technologies is seeking an MLOps Engineer to build and maintain the automation and infrastructure for developing, deploying, monitoring, and managing machine learning models in production.

Why similarSame roleRemote

NVIDIA

1d ago

New

Software Engineer, ML Performance

Full-timeOn-siteSanta Clara, CA$224K – $356K/yr

NVIDIA is hiring a Software Engineer, ML Performance to optimize performance across distributed AI training and inference workloads.

Why similarMachine Learning & AICUDASanta Clara

Apple

1d ago

New

Senior Computer Vision & Machine Learning Engineer

Full-timeOn-siteSunnyvale, CACompetitive

Apple is seeking a Senior Computer Vision & Machine Learning Engineer to develop real-time 3D perception, spatial input, and tracking technologies for future products.

Why similarMachine Learning & AIC++Sunnyvale

Get new Machine Learning, Model Training and Inference jobs in San Francisco by email

Create job alert

Research Scientist (Generative Modeling)

AMD · $150K – $215K/yr

Apply on company site