Robotics Engineer, Perception

About Contoro

Contoro Robotics is an Austin-based startup focused on warehouse automation. We design a state-of-the-art autonomous truck unloading system capable of lifting boxes over 60 lbs.

Role

We are hiring a robotics engineer to maximize the accuracy of our box detection and singulation. You will own the machine learning pipeline that turns raw sensor data into reliable box detections - model training, dataset curation, evaluation, and edge deployment - working alongside our perception engineers to push detection and singulation accuracy across the full range of box sizes and container conditions we see in production.

Responsibilities

• Train, evaluate, and deploy instance segmentation and detection models that improve box detection and singulation accuracy, including for small, occluded, deformed, and tightly-packed boxes
• Build and maintain automated dataset curation and ground-truth generation pipelines, including foundation-model-assisted labeling (e.g., SAM) to scale training data
• Own a deterministic benchmarking and regression framework that evaluates model performance across real and simulated datasets, stratified by box size, container type, and failure mode
• Optimize models for real-time inference on edge hardware using TensorRT and quantization, balancing accuracy against latency and memory budgets
• Debug and resolve production detection failures through log analysis, failure-case review, and targeted retraining
• Collaborate with perception engineers on calibration, localization, and the interface between detections and downstream planning
• Participate in design reviews and contribute to module-level technical decisions

Qualifications

• B.S. or M.S. in Computer Science, Robotics, Electrical Engineering, or a related field
• 3+ years of professional experience developing and deploying computer vision / ML models for real-world systems
• Proficiency in Python and PyTorch in a production environment; working knowledge of C++
• Hands-on experience with instance segmentation or object detection models (e.g., Mask R-CNN, Detectron2, YOLO, SAM)
• Experience building dataset curation, labeling, or evaluation pipelines
• Experience deploying models to edge hardware (NVIDIA Jetson or similar) with TensorRT or comparable inference optimization
• Strong debugging skills and the ability to diagnose model and pipeline failures in production
• Familiarity with Linux-based development environments and ROS / ROS2

PREFERRED QUALIFICATIONS

• Experience with 3D perception and point cloud processing (PCL, Open3D) alongside 2D detection
• Experience with multi-sensor (camera + LiDAR) calibration and synchronized data pipelines
• Experience with stratified model evaluation and regression testing for ML systems
• Familiarity with Docker-based deployment and cloud-based logging/monitoring
• Prior work in warehouse automation, logistics, or pick-and-place applications