Staff AI Software Engineer - Edge Model Optimization & Deployment

FieldAI • Irvine, CA

2d

About The Position

Field AI builds field-proven embodied AI that enables robots to operate autonomously in complex, unstructured real-world environments. Our systems perceive, reason, and act directly on the robot, running on edge hardware under strict constraints on latency, power, and reliability.

We focus on translating cutting-edge AI research into deployable, production-grade autonomy, with an emphasis on robustness, efficiency, and real-world performance. Our AI stack runs on embedded platforms such as NVIDIA Jetson and Orin, powering robots that operate continuously without reliance on cloud compute or curated environments.

Requirements

• 5+ years of professional experience developing and deploying deep learning models for edge, embedded, or real-time systems.
• BS, MS, PhD, or equivalent experience in Computer Science, Robotics, Electrical/Computer Engineering, or a related field.
• Strong proficiency in PyTorch, C++, Python, and CUDA for AI/ML development and model optimization.
• Hands-on experience with TensorRT, ONNX, and Triton, including authoring custom plugins for TensorRT.
• Proven experience applying model optimization techniques such as quantization, pruning, and distillation in production systems.
• Deep understanding of hardware constraints and performance tuning on Jetson / ARM platforms, GPUs, and embedded Linux systems.
• Experience integrating AI models into ROS-based robotic systems.
• Ability to work independently while collaborating effectively in a fast-paced, cross-functional engineering environment.

Nice To Haves

• Experience with ROS2.
• Experience writing and optimizing custom CUDA kernels and low-level GPU performance tuning.
• Familiarity with Triton, ML compilers, or compiler-level optimizations for GPU inference.
• Experience with JAX or additional ML frameworks beyond PyTorch.
• Background deploying AI systems on real robots operating in the field, not just offline or in simulation.
• Familiarity with NVIDIA’s edge and robotics ecosystem (e.g., Isaac ROS, DeepStream, JetPack).

Responsibilities

• Convert and optimize 2D/3D CNNs and Transformer-based models (PyTorch/TensorFlow → ONNX → TensorRT/Triton) for real-time inference on Jetson/Orin platforms.
• Apply model compression techniques—quantization, pruning, distillation, weight sharing—to meet strict constraints on latency, memory, bandwidth, and power.
• Develop custom TensorRT plugins and CUDA kernels for performance-critical components.
• Integrate optimized models into the broader robotic system using ROS nodes and interfaces.
• Build benchmarks, profile and debug end-to-end inference pipelines, and validate performance in real-world robotic scenarios.
• Collaborate closely with AI researchers, robotics engineers, and hardware teams to translate cutting-edge research into robust, deployable edge solutions.
• Ensure the reliability, robustness, and stability of deployed models operating continuously in challenging, resource-constrained environments.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level