About Maincode

Maincode is an AI research lab building systems that move humans forward. We are training foundation models from scratch, designing new reasoning architectures, and deploying them on state-of-the-art GPU clusters. Recently, we’ve built Matilda, Australia’s first LLM. This is not fine-tuning someone else’s work. We’re building new systems from first principles.

Our core research focuses on the development of foundation models and novel architectures that underpin reasoning, alignment and generalisation. Our residency program provides a space for researchers to explore complementary and experimental areas of AI, such as:

• Non-standard language model architectures and techniques
• Multi-modal systems, e.g. vision-language models
• Advanced multi-agent system dynamics
• Reinforcement learning for reasoning or planning
• Optimisation of inference or distributed training stack
• Alignment and interpretability techniques

If you are driven by curiosity, motivated by scientific depth, and excited by the idea of contributing to a lab building models from first principles, you’ll find Maincode an ideal environment to explore ambitious research.

The Role: AI Research Resident

The Maincode Research Residency is a paid 3 to 6 month program designed for advanced AI researchers in Australia. As a resident, you will collaborate with researchers (on-site, hybrid or remote with on-site visits) within our core team to investigate your proposed research project.

Maincode provides the infrastructure and computational resources for you to explore your research ideas at scale. You will work closely with our researchers throughout the process, from refining your proposal and designing experiments, to interpreting results and communicating findings.

The residency provides the opportunity to pursue exploratory, high-impact research that is adjacent to Maincode’s core mission. We encourage projects that push the boundaries of fundamental or applied AI and define the trajectory of future AI systems.

This call is the second round of the AI Research Residency program at Maincode. The successes of the initial round include a hire to the core team as well as publications in the following areas:

• Vision-language reasoning: examining how multi-modal models balance logical inference and perceptual grounding ()
• Video anomaly detection: developing structured prompting frameworks for fine-grained, interpretable reasoning in VLMs ()
• Time series forecasting: using mode-adaptive graph networks for spatio-temporal prediction ()

What You’ll Do

• Propose, refine, and lead a research project that complements Maincode’s mission of building foundational AI systems from first principles
• Design and run experiments at scale, leveraging our compute infrastructure and state-of-the-art training pipelines
• Collaborate closely with Maincode researchers to share knowledge, exchange feedback, and develop new approaches to open problems in AI
• Analyse results rigorously, ensuring that findings are interpretable, reproducible, and impactful
• Share insights through internal discussions, research publications or presentations at AI conferences

Who You Are

• A PhD candidate, postdoctoral researcher or faculty member with deep expertise in machine learning, artificial intelligence, or related fields
• You have a strong track record of publications or ongoing research projects in areas such as deep learning, optimisation, reinforcement learning, probabilistic modelling, or systems-level AI research
• Curious, independent, and self-directed, yet eager to collaborate and learn from others
• Comfortable working with large-scale experiments, distributed training environments, or novel algorithmic ideas
• Excited by ambitious, high-risk/high-reward research that challenges current paradigms

The Process

This application includes a project proposal outlining your area of interest for the residency. Our research team will review your application and proposal, then schedule a follow-up conversation to discuss your ideas and share more about the program.