Efficient Inference Engineer for Visual Models

Hedra; San Francisco (CA); On-site; USD 180,000 - 250,000; Full time

Benefits offered by this job

Competitive compensation and equity

401k

Healthcare (Silver PPO Medical, Vision

Dental)

Lunch and snacks at the office

Job summary

Hedra is seeking an Inference Optimization Engineer to join our San Francisco team, working at the boundary between research and systems to accelerate state-of-the-art visual models at inference time.

You'll profile models, implement optimizations across architectures, runtimes, and kernels, and collaborate with researchers and engineers to push performance while maintaining accuracy. This is a hands-on, hardware-aware role in a fast-moving ML systems group.

Qualifications

• Deep technical expertise in efficient ML inference.
• Strong understanding of how DL models execute on hardware.
• Experience profiling ML workloads and driving performance gains.
• Proficient in Python or C++ for systems work.
• Familiarity with PyTorch, CUDA, or Triton.

Responsibilities

• Work with researchers to make new visual models fast at inference time.
• Profile architectures and workloads to identify bottlenecks.
• Develop approaches to reduce latency and improve GPU utilization.
• Explore optimizations like quantization, sparsity, and kernel fusion.
• Implement GPU kernels and optimize multi-GPU/multi-node execution.
• Build benchmarking and profiling infrastructure to measure impact.

Skills

ML inference

ML systems

GPU computing

Python

C++

Performance profiling

Cross-functional collaboration

Tools

PyTorch

CUDA

Triton

TensorRT

vLLM

SGLang

Job description

Hedra is seeking an Inference Optimization Engineer to join our San Francisco team, working at the boundary between research and systems to accelerate state-of-the-art visual models at inference time.

You'll profile models, implement optimizations across architectures, runtimes, and kernels, and collaborate with researchers and engineers to push performance while maintaining accuracy. This is a hands-on, hardware-aware role in a fast-moving ML systems group.