Roblox
Senior Product Manager, AI Compute Infrastructure
$281K–$331K
San Mateo, CA, United States · Manager · Full-time · Posted Jun 10, 2026
- Experience
- 7+ years of product management experience focused on compute infrastructure or distributed systems at scale.
- Visa sponsorship
- For US-based roles, the company may not be able to employ candidates whose U.S. work authorization is related to certain visa categories, or support future H-1B sponsorship at this time.
About the job
Sets strategy and roadmap for Roblox’s AI compute infrastructure, building reliable GPU and CPU platforms for training, inference, and other workloads across on-premises and cloud environments.
What you'll do
- Drive the Compute Platform roadmap across Kubernetes, managed compute services, distributed systems, and machine fleets.
- Build and scale GPU infrastructure for AI training and inference while improving cost, reliability, and performance.
- Partner with platform teams and cross-functional stakeholders to provide compute primitives and improve platform reliability.
What they ask for
- 7+ years of product management experience focused on compute infrastructure or distributed systems at scale.
- Practical understanding of Kubernetes internals and experience productizing Operators, Controllers, and CRDs.
- Familiarity with GPU/accelerator architecture and scheduling challenges, cloud-native networking, and production-grade compute platforms.
- Experience building compute infrastructure on a cloud provider; ability to balance utilization, latency, cost, delivery speed, and reliability.
Nice to have
- Background in AI model development, training, or inference; kernel-level experience or familiarity with custom kernel drivers.
- Experience building agentic systems for compute or infrastructure.
Pay
$280,540 — $330,950 USD (base salary)
Tools
Managed Kubernetes, Roblox Kubernetes Service, Fleet APIs, GPU, CPU, AWS, Google Cloud Platform (GCP), Azure
AI focus
Build GPU infrastructure for frontier-model training and inference; support AI workloads and improve compute efficiency, reliability, and cost-to-serve.
Model-written summary · the original posting is authoritative