This role pays more than
65%
of similar roles. Most pay
$151,475–$226,350
— the shaded band above.
At the midpoint, this role pays about
$209k
versus about
$189k
for comparable roles.
Based on 240 similar postings.
Employer
About Microsoft
Microsoft Corporation is a global technology leader producing software, hardware, and cloud services including Windows, Office 365, Azure cloud platform, Xbox gaming, and Surface devices. Industry: Software & Cloud Computing
Microsoft currently has
598 open roles
on FindRole.
Listed pay typically runs
$119,800–$234,700
across 586 roles with salary data.
As a Member of Technical Staff, Multimodal Infrastructure, you will join the Microsoft AI team to build large-scale infrastructure supporting the full cycle of multimodal generative model development for products like Copilot. You will collaborate with research scientists and product engineers to design and maintain high-scale data processing pipelines, as well as pretraining and post-training frameworks. Your daily work involves developing inference and serving systems while influencing architectural changes and software roadmaps. The role requires proficiency in C, C++, C#, Java, JavaScript, or Python, alongside experience with PyTorch, Megatron, and Deepspeed. You will utilize distributed data processing tools like Ray and Spark, and implement advanced techniques such as RLHF, quantization, and distillation. This position addresses the technical challenge of building robust infrastructure for multimodal models involving image, video, and audio data.
Design, develop, and maintain large-scale multimodal data processing pipelines.
Build and maintain infrastructure for multimodal model pretraining and post-training frameworks.
Develop and maintain large-scale inference and serving frameworks for multimodal models.
Collaborate with research scientists and product engineers to solve infrastructure-related problems.
Drive architectural changes and influence the roadmap of relevant software and hardware components.
Implement advanced training techniques including RLHF, DPO, and GRPO.
Optimize model performance using quantization, compression, and distillation techniques.
Manage distributed inference scaling across multi-node clusters using tools like Ray Serve and Triton.
What we're looking for
Bachelor's Degree in Computer Science or related technical discipline and 6+ years of engineering experience with coding in languages like C, C++, C#, Java, JavaScript, or Python.
Master's Degree in Computer Science or related field and 8+ years of engineering experience (preferred).
10+ years of engineering experience with coding in C, C++, C#, Java, JavaScript, or Python (preferred).
Proficiency in distributed data processing infrastructure including resource management, fault tolerance, Ray, Spark, and CPU/GPU batch processing optimizations (preferred).
Experience with state-of-the-art model inference and serving frameworks, including vLLM, TensorRT-LLM, SGLang, xDiT, or Cache-DiT (preferred).
Proficiency in deep learning frameworks such as PyTorch, Megatron, and Deepspeed (preferred).
Experience with distributed training techniques like data parallelism, model parallelism, and pipeline parallelism (preferred).
Knowledge of quantization/compression techniques, distillation methods, and reinforcement learning training methods like RLHF or DPO (preferred).
Audio Visual
IPTV
Digital Signage
Cisco
Zoom
Media Signal Transport
infrastructure-as-code
Incident Management
Asset Management
Content Distribution
Streaming
Encoding
Enterprise Security
Configuration Management