Senior Perception Learning Engineer

About the role

Apptronik creates humanoid robots designed to assist humans in logistics, manufacturing, and future domestic or healthcare environments. This role focuses on advancing the perception capabilities of the Apollo humanoid platform to ensure safe and effective interaction with complex surroundings. You will bridge the gap between research and production to help bring these robots to market. The position demands ownership of the full perception stack, from raw sensor inputs to actionable environmental models that drive robot behavior. You will be responsible for ensuring that the Apollo platform can reliably interpret dynamic and unstructured settings encountered in real-world operations. Success in this role requires translating high-level research into robust perception systems that meet stringent safety and reliability standards. You will work closely with cross-functional teams to define requirements and validate that perception solutions meet operational needs across diverse scenarios.

Key facts

What you'll do

• Architect and refine perception and SLAM pipelines, covering tasks like mapping, localization, pose estimation, and scene understanding for human-centric environments.
• Build multi-sensor fusion systems that combine data from LiDAR, cameras, depth sensors, and IMUs to produce coherent and accurate state estimates for mobile manipulation.
• Design and manage data pipelines, training workflows, and inference systems that enable scalable model development, validation, and deployment across robot fleets.
• Research and implement learning-based models for SLAM and 3D scene analysis that address challenges in humanoid manipulation, gait support, and safe navigation.
• Profile perception code paths and monitor telemetry to uphold strict latency and accuracy targets while operating within the thermal and compute limits of edge hardware.
• Coordinate with control, planning, and hardware teams to align perception outputs with motion primitives, grasping strategies, and real-time decision-making processes.
• Utilize synthetic data from simulation platforms such as IsaacSim, combined with carefully curated real-world datasets, to improve model generalization and robustness.
• Define and execute validation benchmarks that quantify perception performance under varying lighting, weather, motion, and occlusion conditions.
• Implement monitoring tools that detect perception failures, support rapid debugging, and provide insights for continuous model improvement in deployed systems.
• Drive the adoption of best practices in software engineering, version control, experiment tracking, and documentation to maintain a high-trust and scalable perception codebase.

Requirements

• Hold an MS or PhD in Robotics, Computer Science, Computer Engineering, or a closely related technical field with a strong focus on perception and learning.
• Bring 3 to 5+ years of hands-on experience developing and shipping SLAM or perception systems intended for real-time vision or robotics applications.
• Demonstrate proficiency in Python and modern C++ along with disciplined software engineering practices such as modular design, testing, and debugging.
• Show deep knowledge of 3D geometry, camera models, and probabilistic estimation techniques including factor graphs, EKF, UKF, VIO, and a variety of SLAM approaches.
• Exhibit expertise in deep learning for computer vision, including segmentation, object detection, multi-object tracking, and 3D perception tasks critical to humanoid operation.
• Experience deploying optimized models onto edge hardware while actively managing thermal, compute, and latency constraints in production conditions.
• Familiarity with AI frameworks such as PyTorch, JAX, or TensorFlow, and vision libraries like OpenCV, Detectron2, or YOLO for rapid prototyping and deployment.
• A track record of solving difficult real-world perception problems where noise, ambiguity, and sensor degradation are common.

Nice to have

• Background in humanoid robotics, bipedal locomotion, or manipulation experience that provides intuition for human-centric perception challenges.
• Experience with established SLAM frameworks such as ORB-SLAM, VINS, Rtabmap, GTSLAM, or Cartographer in demanding environments.
• Skills in classical computer vision, including feature extraction, matching, and geometry-based methods that complement learning approaches.
• Knowledge of model acceleration techniques such as quantization or compression using ONNX Runtime or TensorRT to meet edge deployment targets.
• Familiarity with ROS 2, GStreamer, or zero-copy pipelines that enable efficient data movement and low-latency communication.
• Experience with domain adaptation methods and synthetic data generation strategies that reduce reliance on costly real-world labeling.
• History of contributing to open-source robotics or vision projects that demonstrate collaboration, code quality, and community engagement.

Practical notes

• This is a direct hire position, and outside agency solicitations are not permitted for this role.
• The position requires physical capability to lift up to 15 pounds and to sustain extended periods of desk work in a laboratory or office setting.
• Apptronik is an equal opportunity employer, and all qualified applicants will receive consideration without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, disability, or veteran status.

The Senior Perception Learning Engineer will play a central role in ensuring that Apptronik's humanoid robots can perceive, interpret, and adapt to the world with high reliability. This position sits at the intersection of research, software engineering, and robotics systems, offering the opportunity to influence products that operate alongside humans in diverse professional settings. Candidates should be motivated by challenging technical problems, rigorous engineering standards, and the prospect of deploying perception systems at scale. If you are passionate about building the future of human-aware robotics and want to own the perception stack from algorithm design to edge deployment, this role provides the scope and impact to do so at Apptronik.