What are the responsibilities and job description for the Computer Vision Engineer position at Pittsburgh Robotics Network?
About FuturHand Robotics
FuturHand Robotics is a Carnegie Mellon University spin-off developing next-generation dexterous robotic systems that transform how robots interact with the world. We build on cutting-edge research in dexterous hands and manipulation to solve real-world problems, developing dexterous bimanual robots that operate reliably outside controlled lab environments.
The Role
This role focuses on building the perception systems that enable our robots to understand and interact with the physical world. You will take ownership of the vision pipeline, from raw sensor data through object detection and pose estimation to stable, actionable scene understanding.
You'll work closely with manipulation, controls, and hardware teams to ensure perception outputs are accurate, consistent, and usable for downstream decision-making. This includes defining how objects are represented, how uncertainty is handled, and how the system behaves across a wide range of real-world conditions.
In some areas, you will design new approaches from first principles. In others, you will extend and refine existing pipelines. A key part of the role is building perception systems that are robust, adaptable, and maintainable as the system evolves and new tasks and objects are introduced.
What You'll Be Doing
The most important thing is that you know how to build vision systems that are robust and usable within a larger robotic system. You understand how perception interacts with downstream components and how to design systems that behave predictably under variation and uncertainty.
Beyond That
Email us with your resume and a short note describing a vision system you've built. What did the system need to handle, and what were the hardest parts to get right?
apply@futurhandrobotics.com
FuturHand Robotics is a Carnegie Mellon University spin-off developing next-generation dexterous robotic systems that transform how robots interact with the world. We build on cutting-edge research in dexterous hands and manipulation to solve real-world problems, developing dexterous bimanual robots that operate reliably outside controlled lab environments.
The Role
This role focuses on building the perception systems that enable our robots to understand and interact with the physical world. You will take ownership of the vision pipeline, from raw sensor data through object detection and pose estimation to stable, actionable scene understanding.
You'll work closely with manipulation, controls, and hardware teams to ensure perception outputs are accurate, consistent, and usable for downstream decision-making. This includes defining how objects are represented, how uncertainty is handled, and how the system behaves across a wide range of real-world conditions.
In some areas, you will design new approaches from first principles. In others, you will extend and refine existing pipelines. A key part of the role is building perception systems that are robust, adaptable, and maintainable as the system evolves and new tasks and objects are introduced.
What You'll Be Doing
- Design and implement the full perception pipeline, including sensor integration, object detection, pose estimation, and state tracking
- Develop vision systems that operate reliably across variation in lighting, object geometry, and scene complexity
- Build and maintain system infrastructure for perception: data pipelines, evaluation tools, debugging workflows, and performance metrics
- Work closely with controls and manipulation teams to ensure perception outputs are accurate, stable, and actionable
- Improve robustness through techniques such as sensor fusion, temporal filtering, and data-driven methods
- Develop approaches that generalize across objects and tasks without requiring extensive re-engineering
The most important thing is that you know how to build vision systems that are robust and usable within a larger robotic system. You understand how perception interacts with downstream components and how to design systems that behave predictably under variation and uncertainty.
Beyond That
- Strong fundamentals in computer vision, including 3D perception, object detection, and pose estimation
- Experience working with depth sensors and RGB-D pipelines (e.g., stereo or structured light)
- Experience building and deploying vision systems in real-world or semi-structured environments
- Comfortable working across the stack, from sensor-level considerations to model design and system integration
- MS or PhD in computer vision, robotics, or a related field, or equivalent industry experience
- Based in or willing to relocate to Pittsburgh, PA
- Experience with robotics perception for manipulation tasks
- Familiarity with ROS 2
- Experience with sim-to-real transfer, domain randomization, or data augmentation for robustness
- Experience deploying vision systems on embedded or edge hardware (e.g., Jetson)
- Ownership of the perception system for a novel robotic platform
- The opportunity to shape how robots perceive and interact with complex real-world environments
- A small, highly technical team with close collaboration across disciplines
- Meaningful early-stage equity, competitive salary, and benefits
Email us with your resume and a short note describing a vision system you've built. What did the system need to handle, and what were the hardest parts to get right?
apply@futurhandrobotics.com