Machine Learning Engineer: Perception

Bedrock Robotics; San Francisco (CA); On-site; USD 120,000 - 150,000; Full time

Job summary

Bedrock Robotics is looking for a Machine Learning Engineer in San Francisco, CA, who will focus on developing production 3D perception systems. Candidates should have 3+ years of experience with deep learning models in a production setting, and a understanding of both computer vision and LIDAR approaches. Responsibilities include designing architectures that fuse data, deploying models on embedded hardware, and collaborating on various projects. Strong skills in Python and knowledge of C++ or Rust are essential.

Qualifications

• 3+ years of experience taking deep learning models from research to real-world production.
• Deep understanding of SE(3) transformations and sensor calibration.
• Practical experience with early fusion architectures.
• Expert in Python and familiar with systems code in C++ or Rust.
• Understanding of data alignment and its importance in robotics.

Responsibilities

• Develop and train state-of-the-art models that fuse raw Lidar and Camera data.
• Build robust perception systems for dynamic occlusion and high-vibration conditions.
• Optimize models for inference on embedded hardware.
• Collaborate with other teams for state-of-the-art representations.

Skills

Production ML Experience

3D Geometry & Calibration

Early Fusion Expertise

SOTA Object Detection expertise

Systems Fluency

Data Intuition

Tools

PyTorch

Tensorflow

JAX

C++

Rust

Job description

Machine Learning Engineer: Perception

We are looking for engineers with expertise in shipping production 3D perception systems at scale. Successful candidates have architected systems, trained models from scratch, understand the full stack (clustering, detection, classification, and tracking), and have shipped at scale. We use both computer vision and LIDAR-based approaches, so knowledge of either or both is key. Models are just part of the system: you understand data and have good intuition about why models fail. You know how to evaluate corner cases, manage or build data pipelines, use autolabels (or not), and have a strong understanding of statistical properties of these systems.

What You'll Do

• Design Early Fusion Architectures: Develop and train state-of-the-art models (e.g., BEV-based transformers) that fuse raw Lidar and Camera data to solve for object detection and semantic segmentation.
• Tackle "Messy" Physics: Build perception systems robust enough to handle dynamic occlusion (seeing the robot’s own arm/bucket), particulates (dust, snow, rain), and high-vibration conditions.
• Deploy to the Edge: Optimize models for inference on embedded hardware. You will debug system-level issues, such as sensor calibration drift and latency bottlenecks.
• Collaborating with other teams to create state-of-the-art representations for downstream use cases.

What We're Looking For

• Production ML Experience: 3+ years of experience taking deep learning models from research to real-world production using PyTorch, Tensorflow, or JAX.
• 3D Geometry & Calibration: You have a deep understanding of SE(3) transformations, homogeneous coordinates, and intrinsic/extrinsic sensor calibration. You understand the math required to project a 3D Lidar point onto a 2D image pixel accurately.
• Early Fusion Expertise: Practical experience with architectures that fuse modalities at the feature level (e.g., BEVFusion, TransFuser, PointPainting) rather than just fusing final bounding boxes.
• SOTA Object Detection experience with modern transformer-based architectures (DETR, PETR, etc…) including similar temporal models (PETRv2, StreamPETR, …)
• Systems Fluency: You are an expert in Python, but you are also comfortable reading and writing systems code in C++ or Rust. You understand memory management and real-time constraints.
• Data Intuition: You understand that in robotics, better data alignment often beats a bigger model. You are willing to dig into the data infrastructure to ensure ground truth quality.

Ways To Stand Out

• Bonus: Voxel/Occupancy Experience: Experience working with occupancy grids, NeRFs, or voxel-based representations for terrain mapping.
• Bonus: Top-Tier Research: Published work in conferences such as ICRA, IROS, CVPR, ECCV, ICCV, CoRL, or RSS