Capital One
AI Engineer 5 (Gen AI Platform Services: Agentic AI, Guardrails, Evaluation)
$230K–$286K
San Francisco, CA +4 · Full-time · Posted Oct 7, 2026
- Location
- San Francisco, CA
McLean, VA
Cambridge, MA
San Jose, CA
New York, NY - Experience
- At least 6 years developing AI and ML algorithms or technologies with a bachelor's degree, or at least 4 years with a master's degree; at least 6 years programming. Preferred: 7+ years deploying AI solutions on cloud platforms.
- Education
- Bachelor's or master's degree in Computer Science, AI, Electrical Engineering, Computer Engineering, or related fields.
- Visa sponsorship
- Capital One will consider sponsoring a new qualified applicant for employment authorization for this position.
About the job
Build and lead scalable AI platform systems, including agentic workflows, model orchestration, optimization, guardrails, evaluation, and governance.
What you'll do
- Design, develop, deploy, and support AI components and multi-model orchestration pipelines.
- Optimize foundation model training and inference for scalability, cost, latency, throughput, and hardware utilization.
- Lead design reviews and cost-performance governance, and mentor AI engineers.
What they ask for
- Bachelor's degree in a related field plus at least 6 years developing AI and ML algorithms or technologies, or a master's degree in a related field plus at least 4 years.
- At least 6 years programming with Python, Go, Scala, CUDA, or Java.
Nice to have
- Experience leading AI systems and making tradeoff decisions involving cost, latency, throughput, and accuracy.
- Experience deploying scalable, responsible AI solutions on cloud platforms.
Pay
$229,900 - $262,400 USD base salary; $250,800 - $286,200 USD base salary
Tools
AWS Ultraclusters, Huggingface, VectorDBs, PyTorch, Python, Go, Scala, CUDA
AI focus
Develop agentic AI and multi-agent workflows, foundation model training and inference, model optimization, similarity search, guardrails, evaluation, orchestration, governance, and observability.
Model-written summary · the original posting is authoritative