Simulation and Test Engineer; Conversational AI

About Us

Andromeda Robotics is an ambitious social robotics company with offices in Melbourne and San Francisco, dedicated to creating robots that seamlessly and intelligently interact with the human world. Our first robot, Abi, is a testament to this vision—a custom-built platform designed from the ground up to be a helpful aid and intuitive partner in aged care homes. We are a passionate, collaborative team of engineers who solve some of the most challenging problems in AI and robotics.

To accelerate our development and ensure Abi's reliability, we are seeking a foundational member to build out our capabilities to train and test our robot in simulation.

Deeply empathetic - Kindness and compassion are at the heart of everything we do.

Purposely playful - Play sharpens focus. It keeps us curious, fast and obsessed with the craft.

Relentlessly striving - With relentless ambition, an action bias and constant curiosity, we don’t settle.

Strong when it counts - Tenacious under pressure, we expect problems and stay in motion to adapt and progress.

United in action - Different minds. Shared mission. No passengers.

We are looking for a creative and driven Simulation and Test Engineer to build Andromeda's testing infrastructure for our conversational AI systems and embodied character behaviours. Your immediate focus will be creating robust test systems for Abi's voice-to-voice chatbot, social awareness perception, and gesture motor control. As this infrastructure matures, you'll extend it into simulation environments for generating synthetic training data for character animation and gesture models.

You’ll work at the intersection of our character software, robotics, perception, conversational AI, controls, and audio engineering teams. We bring deep expertise from autonomous vehicles and robotics, including simulation backgrounds. You’ll collaborate with product owners and technical specialists to define requirements, integrate systems, and ensure quality across our AI/ML stack.

• Define and stand up synthetic test environments for our AI/ML conversational stack
• Conversational AI testing: voice-to-voice chat quality, response appropriateness, tool calling accuracy
• Memory system testing: context retention, recall accuracy, conversation coherence
• Audio modelling and testing: multi-speaker scenarios, room acoustics, voice activity detection
• Perception system testing: social awareness (face detection, gaze tracking, person tracking)
• Gesture appropriateness testing:
  Working with our Controls/ML team, create test infrastructure to validate that Abi's body gestures
• CI/CD and automated regression testing for all AI/ML subsystems
• Custodian of quality metrics: if they don't exist, work with stakeholders to elicit use cases, derive requirements, and establish measurable quality metrics
• Requirements formalisation: you're skilled at gathering, documenting, and tracing requirements back to test cases

Our approach to gesture generation requires high-fidelity synthetic interaction data 'll investigate and build the infrastructure to generate this data, working closely with our character software team to define requirements and validate approaches.

• Extend test environments into training data generation pipelines
• Investigate and stand up simulation tools (e.g. Unity, Unreal Engine, Isaac Sim) to support our machine learning pipeline with synthetic data and validation infrastructure
• Build infrastructure for fine-tuning character animation models on simulated multi-actor scenarios
• Enable ML-generated gesture development to augment hand-crafted animation workflows
• Create virtual environments with diverse social interaction scenarios for training and evaluation

Months 1-3, stabilise our conversational system with automated regression tests and measurable quality benchmarks. By month 6, deliver an integrated simulation environment enabling rapid testing and iteration across our AI/ML stack.

You’ll design tests that push our systems beyond their limits and find what's brittle. Through trade studies and make-vs-buy…