Research Scientist

Company Description Metacognition is a NeoLab focused on building AI that enables teams of robots to operate safely and productively in complex, real-world environments. The company’s researchers design new architectures and theories that go beyond transactional language models, creating systems that can manage competing goals, long time horizons, and continuous adaptation. Building on foundational work in Vision and Language Navigation, Metacognition develops AI for ongoing real-time systems such as robots, mines, and power grids. We're building AI that learns from experience, questions its own assumptions, and is robust to prompt injection. Founded by leading AI researchers and commercialization experts from AIML, Amazon, and Stanford, Metacognition is creating the missing software layer that allows digital agents and physical robots to be trained by everyday people through simple instruction and feedback.

Role Description These are full-time remote roles based in Australia. The Research Scientist will design and analyze novel AI architectures for real-time robotic and industrial systems, with a focus on learning from experience, and robust decision-making. Day-to-day activities include formulating research hypotheses, developing and implementing algorithms, running experiments and simulations, and rigorously evaluating performance on complex tasks in general environments. The role involves close collaboration with engineering and product teams to translate research insights into deployable systems, as well as contributing to publications, internal technical reports, and open-source or shared research artifacts. The Research Scientist will also engage in code reviews, participate in technical discussions, and help shape the roadmap for next-generation AI that can safely control robots and other critical infrastructure.

Qualifications

• Enthusiasm for the latest agentic programming methods. We're an AI-first company.
• A PhD in AI (or a related area) and a good publication record are highly rated, particularly for the research positions, but not essential for the engineering focussed roles.
• Strong foundation in machine learning and AI, including experience with deep learning, reinforcement learning, and sequential decision-making.
• Expertise in areas such as robotics, autonomous systems, or operating systems (yes, operating systems), with emphasis on real-time or safety-critical applications.
• Proficiency in programming languages commonly used in ML research (e.g., Python, C++), and familiarity with ML frameworks (e.g., PyTorch, TensorFlow, JAX).
• Experience designing and conducting empirical research: problem formulation, experimental design, data analysis, and interpretation of results.
• Ability to work with large-scale or complex systems, including simulation environments, multi-agent setups, or infrastructure-level applications.
• Strong written and verbal communication skills for collaborating across disciplines and documenting research for internal and external audiences.
• PhD or equivalent research experience in Computer Science, Robotics, Electrical Engineering, Applied Mathematics, or a related field.
• Experience with language models, vision-and-language systems, or metacognitive architectures is highly beneficial.
• Comfort working in a fast-paced, experimental environment. We're moving fast and making things. Breaking a few things is ok too, but we're focussed on the making.