| Microsoft Careers

What you'll do

• Design and lead development of high-performance ad-serving systems with ultra-low latency.
• Optimize end-to-end inference infrastructure for cost efficiency across heterogeneous hardware.
• Profile and optimize full-stack performance, identifying and resolving bottlenecks in real-time.
• Own live-site reliability by designing telemetry and fault-tolerance mechanisms globally.
• Mentor team members on deep debugging, profiling, and system-level optimization practices.

What we're looking for

• Extensive experience in designing and optimizing high-performance ad-serving systems.
• Hands-on expertise with real-time data streaming systems and multi-region deployment.
• Deep knowledge of LLM inference optimization techniques including model sharding and quantization.
• Proven success in operating large-scale systems with SLA-based capacity forecasting and autoscaling.
• Expertise in GPU inference frameworks like NVIDIA Triton Inference Server, CUDA, and TensorRT.
• Solid understanding of model-serving trade-offs and optimization strategies for deep learning workloads.