Machine Learning Engineer at Syos Aerospace

• Helped design a multi-vehicle, multi-domain uncrewed-systems autonomy stack, and contributed to its DDS-based runtime layer (messaging, logging and IDL service contracts) • Own the perception subsystem: trained and deployed real-time object detectors to Jetson Orin NX as optimised TensorRT engines (30 FPS continuous, at camera rate), and integrated constrained vision-language models for perception tasks • Lifted detection [email protected] from 0.36 (off-the-shelf pretrained baseline) to 0.68 (0.78 precision, 0.61 recall) on a∼261k-image dataset spanning 15 sources; also built GANs for thermal/RGB fusion to strengthen multi-spectral perception • Built multi-sensor contact/data fusion and a coordinated awareness engine performing fleet-wide track fusion for shared situational awareness • Developed a COLREGS compliance agent for safe surface navigation, and contributed to route planning, geo-intelligence and guidance • Contributed to an LLM-based natural-language mission-intent system that pairs LLM segmentation and slot-extraction (local Ollama or OpenAI-compatible) with deterministic validation, a multi-turn confirmation state machine and human-gated outputs (Python, Pydantic v2, LangChain) • Currently helping build on-vehicle cognition (decision-making) and fleet management; instrumented ML training pipelines with MLflow (experiment tracking, model registry) and drift monitoring • Guide two junior ML engineers and lead the company’s ML/AI work, partnering with the Head of Software & Autonomy on system design and architecture

Software & Artificial Intelligence Engineer at Marine AI

Led NZ-side deployment, integration and field validation of a UK-developed multi-vehicle autonomy stack, reporting to directors and handling software end-to-end. Developed and deployed real-time perception models for object detection, tracking and recognition across camera and LiDAR in challenging marine conditions, with early work on camera/LiDAR fusion. Built automated model-testing and validation frameworks that improved accuracy-assessment workflows and boosted robustness by fusing labelled and synthetic training data. Prepped and ran on-water sea trials and stakeholder demonstrations validating the integrated stack on real vehicles. Built Python (Flask) APIs to serve perception outputs to downstream components and applied OCR/NLP within broader ML workflows.

Research Student (Honours Project) at Robotics Plus

Completed the Engineering Honours project with Robotics Plus. Advanced Python and C++ skills and used ROS2 and Docker for robotics development. Applied deep learning to improve object detection accuracy. Developed and deployed a tracking system for monitoring detections. Calibrated stereo cameras to generate high-quality point clouds. Implemented a custom test rig to meet project needs. Strengthened code management through proactive Git usage and improved productivity in VS Code and Ubuntu/Linux environments. Conducted research into automation and robotics technologies.

Research Student at University of Waikato

Designed and deployed a vision system for optimal performance on the University’s Rock Melon Harvester. Explored research topics and produced detailed summaries while deepening Python and C++ proficiency. Built Docker-based isolated workspaces to support ROS2 development. Implemented neural networks to improve detection accuracy and added tracker, sizing algorithm, occlusion checks, and segmentation techniques. Applied OCR for text extraction within the vision pipeline. Created a robust outdoor-capable rig handling weather and lighting variability. Improved code documentation and management practices, and orchestrated Git server setup for efficient version control.

Project Engineering Intern at Robotics Plus

Applied mechanical, electrical, and software expertise to assemble robotic systems and support installations. Performed camera calibration, system integration, and electrical component assembly. Streamlined robotics assembly processes by improving documentation, workflows, and production models. Managed multiple concurrent projects while supporting robot assembly and troubleshooting. Developed strong collaboration, organizational, and time management skills in production environments.