Staff ML Infra Engineer: Global, Low-Latency ML Systems

Uber; San Francisco (CA); On-site; USD 232,000 - 258,000; Full time

Benefits offered by this job

Bonus program

Equity award

401(k) plan

Benefits

Job summary

Uber is seeking a Staff Software Engineer for ML Infra in San Francisco to architect and build scalable ML data and backend systems that power the trust and pricing platforms across the global marketplace. You will lead development efforts, mentor engineers, and collaborate with product, science, and ops to translate complex requirements into robust technical roadmaps.

Strong experience in MLOps, large language models, and distributed systems is required, focusing on reliability, low latency,

Qualifications

• 9+ years of experience building ML and backend systems.
• Bachelor's degree in CS/CE or related field.
• Experience with AI/ML frameworks (PyTorch, TensorFlow).
• Experience developing foundational models or LLMs at scale.
• Strong understanding of ML algorithms and infrastructure.
• Proven MLOps and ML infrastructure experience (Kubernetes, pipelines, schedulers).

Responsibilities

• Design and build ML data and backend systems at scale.
• Lead architecture, provide code reviews, and mentor engineers.
• Collaborate with Product, Science, and Ops to roadmap.
• Ensure reliability, low latency, and high availability in production.

Skills

MLOps & ML Infrastructure

Kubernetes & data pipelines

LLMs & foundation models

PyTorch / TensorFlow

Distributed systems & scheduling

Education

Bachelor's degree in CS/CE or related field

Tools

Kubernetes

Data pipelines

Distributed workflow systems

Job description

Uber is seeking a Staff Software Engineer for ML Infra in San Francisco to architect and build scalable ML data and backend systems that power the trust and pricing platforms across the global marketplace. You will lead development efforts, mentor engineers, and collaborate with product, science, and ops to translate complex requirements into robust technical roadmaps.

Strong experience in MLOps, large language models, and distributed systems is required, focusing on reliability, low latency,