Staff AI Engineer: Build Scalable LLM AI Systems

Capital One; New York (NY); On-site; USD 294,000 - 335,000; Full time

Job summary

Capital One is seeking a Distinguished AI Engineer in New York to help build responsible AI systems and groundbreaking products. You will partner with engineers, researchers, PMs, and product managers to deliver AI-enabled solutions that impact associates and customers alike.

You will design and optimize AI software, train foundation models, perform inference, and tackle governance, observability, and cost challenges across a broad tech stack including AWS, PyTorch, and VectorDBs.

Qualifications

• Bachelor's degree with 8+ years AI/ML experience or Master's with 6+ years.
• 8+ years programming in Python, Go, Scala, or Java.
• Experience deploying scalable and responsible AI on cloud platforms.
• Ability to lead and mentor multiple engineering teams.

Responsibilities

• Partner with cross-functional teams to deliver AI-powered products.
• Design, train, test, deploy, and support AI software components.
• Work with Open Source and SaaS AI tech (AWS, Huggingface, PyTorch, etc.).
• Innovate LLM optimization to improve performance, cost, latency, and throughput.
• Contribute to the long-term technical vision and roadmap.

Skills

LLM optimization

Cross-functional collaboration

Communication

Problem solving

Education

Bachelor's degree in Computer Science/AI/Engineering

Master's degree in Computer Science/AI/Engineering

Tools

Python

Go

Scala

Java

PyTorch

AWS

VectorDBs

Guardrails

Job description

Capital One is seeking a Distinguished AI Engineer in New York to help build responsible AI systems and groundbreaking products. You will partner with engineers, researchers, PMs, and product managers to deliver AI-enabled solutions that impact associates and customers alike.

You will design and optimize AI software, train foundation models, perform inference, and tackle governance, observability, and cost challenges across a broad tech stack including AWS, PyTorch, and VectorDBs.