Perception Engineer

moss robotics; San Francisco (CA); On-site; USD 150,000 - 230,000; Full time

Job summary

moss.inc is seeking a founding engineer to own the development and deployment of our 3D perception pipeline. You will build multimodal models that fuse LiDAR, cameras, GPS, and metadata for robust 3D analysis and autonomous operation.

You will tackle challenging outdoor environments, optimize for real-time edge performance, and contribute to ML infrastructure for labeling, data management, and evaluation. This is a hands-on, fast-moving startup role with real impact.

Qualifications

• Impressive real-world projects beyond the classroom (robotics, perception, mapping, autonomy).
• Hands-on experience with 3D sensor data (LiDAR, radar, depth cameras).
• Strong C++ (templates, smart pointers, STL containers, algorithms).
• Experience training ML models on custom datasets (data curation, training/eval loops).
• Experience with object detection, semantic segmentation, and/or classical CV / point-cloud methods (e.g., clustering, registration, tracking).

Responsibilities

• Own the full lifecycle of our 3D perception pipeline (research, prototype, production, deployment, iteration).
• Build multimodal models and pipelines that fuse LiDAR, cameras, GPS and metadata for 3D detection and analysis.
• Design robust algorithms for outdoor environments (harsh shadows, lighting shifts, motion blur, dust, severe occlusions).
• Help build and maintain ML infrastructure for automated labeling, dataset management, training, and evaluation.
• Optimize and deploy models for real-time performance on edge hardware (latency, throughput, memory).
• Explore new approaches like vision-language-action (VLA) models, imitation learning, and autonomy-oriented perception for robot tasks.

Skills

3D sensor data

C++

ML models training

Object detection

Semantic segmentation

Point-cloud methods

Robotics perception

Autonomy

Job description

About Us

At moss.inc, we are making the physical world searchable. Our sensor kits mount to vehicles our customers already drive, and turn every drive into a measured 3D record of their site, across agriculture, construction, and energy. Our AI models then parse through those 3D maps and decipher important information for each of our customers. We started with tree farms, where a single field holds millions of plants no human has ever fully inventoried.

We're a small team of practical engineers with a long-term vision. We focus on real, messy, on-the-ground problems today, while working toward a future where AI agents understand the physical world.

If our mission aligns with how you work and think, we'd love to learn more about you.

The Role

Join us as a key founding engineer. We’re a small, fast-moving team of 7, developing novel perception systems and algorithms to interpret the physical world in challenging, real-world environments.

You will own the development and deployment of our 3D perception pipeline. This involves developing novel algorithms and multi-modal models (LiDAR, Camera, GPS & Environmental Data) for understanding farms and enabling autonomous robot operations.

We move quickly, solve hard problems, and are excited by the reactions we receive from our customers.

We're looking for both full time (in-person) roles and intern candidates for co-ops, summer, and/or part-time.

Minimum Requirements

• Impressive real-world projects beyond the classroom (robotics, perception, mapping, autonomy, etc.)
• Hands-on experience with 3D sensor data (LiDAR, radar, depth cameras)
• Strong C++ (templates, smart pointers, STL containers, algorithms)
• Experience training ML models on custom datasets (data curation, labeling, training/eval loops)
• Experience with object detection, semantic segmentation, and/or classical CV / point-cloud methods (e.g., clustering, registration, tracking)

What You’ll Do

• Own the full lifecycle of our 3D perception pipeline (research, prototype, production, deployment, iteration)
• Build multimodal models and pipelines that fuse LiDAR, cameras, GPS and metadata for 3D detection and analysis
• Design robust algorithms for outdoor environments (harsh shadows, lighting shifts, motion blur, dust, severe occlusions)
• Help build and maintain ML infrastructure for automated labeling, dataset management, training, and evaluation
• Optimize and deploy models for real-time performance on edge hardware (latency, throughput, memory)
• Explore new approaches like vision-language-action (VLA) models, imitation learning, and autonomy-oriented perception for robot tasks