Robotics Engineer

ProFound People; Melbourne; Remoto; AUD 90.000 - 150.000; Tempo pieno

Descrizione del lavoro

ProFound People Pty Ltd in Australia is seeking a Robotics Engineer to lead learning-based robot behaviour for industrial construction. You will turn demonstrations into dependable robot actions, shape data collection and learning approaches, and spend significant time with hardware on-site and in the lab.

The role blends practical robotics with learning methods, requiring 2+ years of commercial experience and strong Python and deep learning skills.

Competenze

• At least 2 years of full-time commercial experience getting learned behaviour working on physical robots.
• Solid Python programming and experience with deep learning training workflows.
• Experience with learning from demonstration or reinforcement learning applied to robot control.
• Experience using physics-based simulation to generate training data for robots.
• Hands-on time operating, calibrating and troubleshooting robot arms.
• Evidence of carrying a technical project from concept through to a working result.

Mansioni

• Gather robot demonstration data, including by remotely operating robotic arms, to feed the training process.
• Develop and refine learned control policies for manipulation and assembly tasks.
• Run structured evaluations to measure policy performance and guide improvements.
• Create and extend simulated training environments with varied scenes for robustness.
• Investigate field failures and implement fixes so policies work in real hardware.
• Contribute to long-term toolchain decisions for the learning work.

Conoscenze

Python

Deep learning

Reinforcement learning

Robot control

Robot simulation

C++

Formazione

Bachelor's degree in a relevant field

Descrizione del lavoro

A young Australian robotics business putting learning based robots to work on industrial construction projects is looking for a Robotics Engineer to lead how those robots learn new tasks.

In this role you will be responsible for turning manipulation and assembly tasks into behaviour a robot arm can perform dependably outside the lab. That means shaping how demonstration and simulated data is gathered, choosing and refining the learning approach, and spending real time with the hardware working out why a behaviour that looked solid in training breaks down in the field. The team is small and you would join near the beginning, so your judgement will set the direction of the learning work rather than slot into an existing plan, and you will see your results running on active construction sites. It suits someone who enjoys the practical side of robot learning as much as the modelling, from calibrating a setup to rerunning a failed trial until it works. This is an industry role rather than research one, and the background that fits best includes at least 2 years of full-time commercial experience getting learned behaviour working on physical robots.

This is a full-time position that can be worked remotely from anywhere in Australia, with periodic visits to project sites when new capability is being trialled.

Key Responsibilities:

• Gather robot demonstration data, including by remotely operating robotic arms, to feed the training process.
• Develop and refine learned control policies that allow robot arms to complete manipulation and assembly tasks.
• Run structured evaluations to measure policy performance and decide where improvement effort should go.
• Create and extend simulated training environments, varying scenes and conditions to make policies more robust.
• Investigate failures seen on physical hardware and make the changes needed for policies to behave in the field as they do in simulation.
• Contribute to longer term decisions on the tools, methods and architecture behind the learning work.
• Visit project sites periodically to trial and commission new robot capability.

Requirements:

• At least 2 years of full time commercial experience, beyond academic or laboratory research, getting learned behaviour working on physical robots.
• Solid Python programming and experience with deep learning training workflows.
• Working knowledge of learning from demonstration or reinforcement learning applied to robot control.
• Experience using physics-based simulation to produce training data for robots.
• Practical time spent operating, calibrating and troubleshooting robot arms.
• Evidence of carrying a technical project from early concept through to a working result.

Desirable but not essential:

• Publicly visible work in robot learning, such as code repositories, papers or personal builds.
• Time spent on construction, electrical or industrial sites.
• C++ programming experience.
• Familiarity with world modelling approaches to robot learning.