Research Scientist - AI-Enabled Decision Making

About Praetorian Aeronautics

Praetorian Aeronautics specialises in developing advanced autonomous aerial platforms aimed at combating the swiftly evolving challenges of drone warfare. With headquarters in Adelaide and offices in Melbourne and Darwin, the company offers a comprehensive ecosystem of counter-drone autonomy systems. These include high-speed interceptors for kinetic drone neutralisation and AI-driven command and control technologies enabling operators to deploy interceptors at scale efficiently. Together, these capabilities allow defence personnel to detect, evaluate, and neutralize autonomous threats while maintaining situational advantage in hostile environments.

Role Overview

The position focuses on defining how intelligent systems reason, plan, and act amid uncertainty—crucial for mission systems countering coordinated swarm threats. Before an autonomous interceptor is deployed, understanding the situational context is vital, such as determining which system targets specific threats and how interceptors coordinate with other assets under data limitations and strict communication and resource constraints. These are complex decision-making challenges that require explanations understandable to human operators.

As a research scientist, you will collaborate closely with flight sciences and software engineering teams to design and evaluate decision-making algorithms tailored for autonomous and semi-autonomous operations. Your work will involve framing challenges as Markov decision processes (MDPs) or partially observable MDPs (POMDPs), applying and extending methods such as reinforcement learning, Monte Carlo tree search, value or policy iteration, alongside combinatorial optimisation techniques addressing resource allocation and task scheduling within large and structured action spaces.

Key Responsibilities

• Formulate sequential decision-making issues as MDPs or POMDPs using reinforcement learning, Monte Carlo tree search, and value or policy iteration.
• Develop combinatorial optimisation solutions—including linear and integer programming, genetic algorithms, and heuristic methods—for both single and multi-objective resource allocation and scheduling.
• Prototype algorithms using Python and standard machine learning/deep learning tools such as Scikit-Learn and PyTorch, then validate these against real-world operational scenarios.
• Enhance decision-making frameworks to handle partially observable and multi-agent environments where applicable.
• Collaborate with flight sciences and engineering teams to transition research prototypes into deployable, operational systems.
• Contribute to academic and industry publications that align with organisational objectives.

Required Qualifications and Experience

• PhD in Machine Learning or a closely related discipline, preferably complemented by additional research experience such as postdoctoral work, industry research, or ongoing publications.
• Solid expertise in MDP-based decision-making methods including reinforcement learning, Monte Carlo tree search, or value/policy iteration techniques.
• Proven background in combinatorial optimisation approaches covering linear/integer programming, genetic algorithms, and greedy heuristics for single and multiple objectives.
• Strong programming skills in Python with proficiency in machine learning and deep learning libraries like Scikit-Learn and PyTorch.

Preferred Skills

• Experience working with partially observable or multi-agent decision-making problems.
• Knowledge of C++ or Rust programming languages.
• Familiarity with large language models (LLMs).
• Background in defence or military application domains.
• Experience in path planning algorithms.

Additional Information

Due to the sensitive nature of the work, applicants must be eligible to work in Australia and hold citizenship from Australia or another country part of the Five Eyes intelligence alliance (Australia, United States, United Kingdom, Canada, New Zealand). This role does not require formal security clearance.