Staff Machine Learning Engineer - Perception

Fruition Group US; San Jose (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Fruition Group US is seeking a Staff-level ML perception engineer to own perception for a production autonomous trucking stack. You will shepherd models from architecture to deployment on trucks hauling real freight, including night driving and construction zones.

You will design perception models, build motion forecasting, and lead training pipelines. This role requires shipping ML-based perception on safety-critical hardware and mentoring other engineers in a fast-paced autonomous-vehicle

Qualifications

• Experience shipping ML-based perception in safety-critical hardware.
• Demonstrated ability to own models from research to deployment.
• Strong Python and ML framework proficiency.

Responsibilities

• Design and own perception models for object detection/segmentation/3D scene understanding.
• Build motion forecasting models for dynamic actors under real traffic conditions.
• Develop and improve training pipelines, datasets, and evaluation frameworks.
• Optimize models for real-time inference on automotive-grade compute.
• Close the sim-to-real gap; handle edge cases like night driving and construction zones.
• Set technical direction for the perception/ML stack; mentor engineers.
• Collaborate with planning, controls, and safety teams to deploy perception in production.

Skills

CV/Deep Learning

Sensor Fusion

Transformer Architectures

Python

ML Frameworks (PyTorch)

Real-time Inference

Embedded/Automotive Compute

Autonomous Vehicles domain

Mentorship

Tools

PyTorch

CUDA

NVIDIA DRIVE

ML Pipelines

Job description

The Role

We're looking for a Staff-level engineer to own ML-driven perception for a production autonomous trucking stack; from model architecture through deployment on trucks hauling real commercial freight, where our autonomous driving software powering self-driving commercial trucks on long-haul highway routes, including night driving and construction-zone handling.

Responsibilities

• Design and own perception models for object detection, segmentation, and 3D scene understanding at highway speeds and long sensing ranges.


• Build motion forecasting models that predict the trajectories of dynamic actors — merging vehicles, lane changes, pedestrians — under real traffic conditions.


• Build and improve training pipelines, datasets, and evaluation frameworks, including transformer-based architectures for perception and forecasting.


• Optimize models for real-time inference on automotive-grade, multi-node compute platforms.


• Close the sim-to-real and lab-to-field gap, hardening models against edge cases like night driving, construction zones, and adverse weather.


• Set technical direction for the perception/ML stack and mentor other engineers.


• Partner with planning, controls, and safety teams to deploy perception in a production, safety-critical system.

Required Skills

• Deep expertise in computer vision, deep learning, or sensor fusion, including transformer-based model architectures.


• Track record shipping ML-based perception on real, safety-critical hardware — not just benchmarks.


• Strong Python and ML framework proficiency (PyTorch or similar).


• Experience with data pipelines, large-scale model training infrastructure, and evaluation at scale.


• Experience optimizing models for real-time inference on embedded or automotive-grade compute (e.g., NVIDIA DRIVE-class platforms).


• Experience with autonomous vehicles or long-haul/highway driving domains specifically.


• Multi-sensor fusion (camera, radar, LiDAR) experience at extended sensing range.


• Experience with motion forecasting or trajectory prediction for dynamic traffic actors.