“I enjoy research, but I’m at my best when I’m making it work.”

“I want more ownership than maintaining one small part of a mature platform.”

“I like building the infrastructure that helps good researchers move faster.”

“I want difficult engineering problems without losing contact with the models.”

If any of this resonates, please keep reading.

An early-stage Australian AI research company is building intelligent systems designed to keep working, learning and adapting over long periods.

That creates a different engineering problem from serving a model behind an API.

State needs to survive between interactions. Work needs to pause, resume and respond to changing priorities. Multiple AI processes need to share resources without getting in each other’s way. New research needs to move from an experiment into something reliable enough to test in real environments.

This team is developing the models and underlying infrastructure required to make that possible.

The research spans memory, reasoning, continual learning and model adaptation. Alongside it, the engineering team is building the training, evaluation and runtime systems needed to turn those ideas into working technology.

This is not a conventional production ML role, and it is not about connecting existing models to another application.

You’ll work at the point where experimental research becomes reliable software.

Reporting to the Chief Engineer, you’ll collaborate closely with research scientists and other engineers to build the systems that allow new models and architectures to be trained, evaluated, demonstrated and eventually deployed.

Day to day

• Turn experimental research code into reliable, efficient and maintainable systems.
• 
  
• Implement and optimise machine learning models and training pipelines.
• 
  
• Build data, simulation and evaluation infrastructure for new research.
• 
  
• Profile model behaviour and performance across different compute environments.
• 
  
• Improve the reliability, scalability and repeatability of ML experiments.
• 
  
• Build demonstrators that test research ideas in realistic settings.
• 
  
• Support cloud infrastructure and the services used by the research and engineering teams.
• 
  
• Work on distributed, real-time or embedded capabilities as the platform develops.
• 
  
• Use AI-assisted development tools while applying strong engineering judgement to the output.
• 
  
• Help shape technical decisions in an environment where many of the answers are still being discovered.
• 
  

Success here will not be measured by how many tickets you close.

It will be measured by whether researchers can move faster, experiments can be trusted and promising ideas can survive the journey from notebook to working system.

Ideal background

• Strong Python engineering skills.
• 
  
• Hands-on experience training and deploying deep learning models using cloud infrastructure.
• 
  
• Commercial or research experience with PyTorch or JAX.
• 
  
• Experience building dependable ML training, data or evaluation pipelines.
• 
  
• An understanding of model performance, compute constraints and distributed workloads.
• 
  
• Experience with large-scale model training, including language or vision-language models.
• 
  
• Familiarity with simulation environments, synthetic data or evaluation systems.
• 
  
• Experience in a research lab, deep-tech company or ambitious AI startup.
• 
  
• Evidence that you have taken experimental work and turned it into something other people could use.
• 
  
• Comfort working independently, making decisions with incomplete information and changing direction as the technology develops.
• 
  

A PhD would be useful, particularly if it came with strong implementation experience, but it is not essential.

This probably won’t suit someone looking for a mature platform, a tightly defined backlog or several layers of product management between them and the problem.

It may suit someone who wants to work directly with strong researchers, influence an emerging technical architecture and build systems that do not already have an established playbook.

• The team is distributed and meets in person regularly. Australia is preferred, although exceptional international candidates will be considered.

“I enjoy research, but I’m at my best when I’m making it work.”

“I want more ownership than maintaining one small part of a mature platform.”

“I like building the infrastructure that helps good researchers move faster.”

“I want difficult engineering problems without losing contact with the models.”

If any of this resonates, please keep reading.

An early-stage Australian AI research company is building intelligent systems designed to keep working, learning and adapting over long periods.

That creates a different engineering problem from serving a model behind an API.

State needs to survive between interactions. Work needs to pause, resume and respond to changing priorities. Multiple AI processes need to share resources without getting in each other’s way. New research needs to move from an experiment into something reliable enough to test in real environments.

This team is developing the models and underlying infrastructure required to make that possible.

The research spans memory, reasoning, continual learning and model adaptation. Alongside it, the engineering team is building the training, evaluation and runtime systems needed to turn those ideas into working technology.

This is not a conventional production ML role, and it is not about connecting existing models to another application.

You’ll work at the point where experimental research becomes reliable software.

Reporting to the Chief Engineer, you’ll collaborate closely with research scientists and other engineers to build the systems that allow new models and architectures to be trained, evaluated, demonstrated and eventually deployed.

Day to day

• Turn experimental research code into reliable, efficient and maintainable systems.
• 
  
• Implement and optimise machine learning models and training pipelines.
• 
  
• Build data, simulation and evaluation infrastructure for new research.
• 
  
• Profile model behaviour and performance across different compute environments.
• 
  
• Improve the reliability, scalability and repeatability of ML experiments.
• 
  
• Build demonstrators that test research ideas in realistic settings.
• 
  
• Support cloud infrastructure and the services used by the research and engineering teams.
• 
  
• Work on distributed, real-time or embedded capabilities as the platform develops.
• 
  
• Use AI-assisted development tools while applying strong engineering judgement to the output.
• 
  
• Help shape technical decisions in an environment where many of the answers are still being discovered.
• 
  

Success here will not be measured by how many tickets you close.

It will be measured by whether researchers can move faster, experiments can be trusted and promising ideas can survive the journey from notebook to working system.

Ideal background

• Strong Python engineering skills.
• 
  
• Hands-on experience training and deploying deep learning models using cloud infrastructure.
• 
  
• Commercial or research experience with PyTorch or JAX.
• 
  
• Experience building dependable ML training, data or evaluation pipelines.
• 
  
• An understanding of model performance, compute constraints and distributed workloads.
• 
  
• Experience with large-scale model training, including language or vision-language models.
• 
  
• Familiarity with simulation environments, synthetic data or evaluation systems.
• 
  
• Experience in a research lab, deep-tech company or ambitious AI startup.
• 
  
• Evidence that you have taken experimental work and turned it into something other people could use.
• 
  
• Comfort working independently, making decisions with incomplete information and changing direction as the technology develops.
• 
  

A PhD would be useful, particularly if it came with strong implementation experience, but it is not essential.

This probably won’t suit someone looking for a mature platform, a tightly defined backlog or several layers of product management between them and the problem.

It may suit someone who wants to work directly with strong researchers, influence an emerging technical architecture and build systems that do not already have an established playbook.

• The team is distributed and meets in person regularly. Australia is preferred, although exceptional international candidates will be considered.