Senior ML Scientist - Inference & Hardware Acceleration

Netskope; Santa Clara (CA); On-site; USD 182,500 - 260,500; Full time

Job summary

Netskope is seeking a Senior Staff Machine Learning Scientist in Santa Clara to own the inference and optimization layer for AI in agentic workflows. You will fine-tune models, push latency and throughput on real hardware, and build a runtime that executes bounded AI tasks with real customer data signals.

You will work on quantization, KV-cache optimization, and hardware acceleration, partnering with systems and backend engineers to ship end-to-end capabilities in production environments.

Qualifications

• MS or PhD in a technical field with focus on AI/ML research.
• 10+ years of industry experience with 4+ years in ML/AI roles.
• Experience with model fine-tuning and inference optimization.
• Strong Python and capability to interface with C++ for low-level work.
• Experience with transformer models and production-grade inference.

Responsibilities

• Build and optimize the model inference path with quantization and KV-cache optimization.
• Fine-tune models and develop evaluation harnesses for real accuracy and latency.
• Design and grow the bounded task execution runtime for scalable workflows.
• Pave the path for hardware acceleration and sparsity for larger models.
• Collaborate with systems and backend engineers to ship end-to-end capabilities.

Skills

Python

C++ interop

Fine-tuning

Quantization

Inference runtimes

Transformers

KV cache

Hardware acceleration

Agentic AI tools

Education

MS in Computer Science / ML / EE

PhD in a related field (preferred)

Tools

vLLM

ONNX Runtime

TensorRT-LLM

llama.cpp

MLX/CoreML

Job description

Netskope is seeking a Senior Staff Machine Learning Scientist in Santa Clara to own the inference and optimization layer for AI in agentic workflows. You will fine-tune models, push latency and throughput on real hardware, and build a runtime that executes bounded AI tasks with real customer data signals.

You will work on quantization, KV-cache optimization, and hardware acceleration, partnering with systems and backend engineers to ship end-to-end capabilities in production environments.