Research Associate/ Assistant/ Robotics Engineer in Vision-Language-Action (VLA) Data and Training – Embodied AI

Research Associate/ Assistant/ Robotics Engineer in Vision-Language-Action (VLA) Data and Training – Embodied AI About the RoleWe are seeking a Robotics Research Associate/ Assistant/ Engineer to join our team, where we develop foundation models for general-purpose robotic manipulation. The successful candidate will operate at the intersection of two complementary functions: the design and execution of high-quality robotic data-collection system, and the training, fine-tuning, and rigorous evaluation of state-of-the-art VLA models trained on the resulting datasets. The position is well suited to a recent graduate or early-career researcher who has demonstrable, hands-on experience with both physical robotic systems and modern multimodal deep learning frameworks. Candidates who are passionate about closing the loop between perception, language, and physical action are particularly encouraged to apply. Key ResponsibilitiesData Collection and Curation•        Design, plan, and execute teleoperation-based data-collection campaigns across a range of manipulation tasks, including defining task taxonomies, configuring scenes, and supervising data-collection sessions. •        Operate, maintain, and troubleshoot robotic manipulation platforms (e.g., Franka Emika, UniversalRobots, xArm, WidowX, or comparable systems) in a laboratory environment. •        Establish and enforce standards for dataset quality, including episode segmentation, natural-language annotation, action-space normalization, and curation. Model Training and Evaluation•        Fine-tune and adapt state-of-the-art VLA architectures (e.g., OpenVLA, Octo, Pi-0, RT-2-class models) and design new models on proprietary and open datasets. •        Design and execute on-robot evaluation protocols, including the quantitative assessment of task success rates, characterization of failure modes, and analysis of out-of-distribution generalization. •        Contribute to the iterative refinement of model architectures, training recipes, and data-augmentation strategies based on empirical evaluation findings. Research and Knowledge Sharing•        Monitor relevant academic literature in vision-language modelling, robot learning, and multimodal foundation models, and translate emerging methods into prototypes and production systems. •        Document methodologies, results, and lessons learned to support reproducibility and knowledge transfer within the team. Required Qualifications•        A Bachelor’s or Master’s degree(completed or in progress) in Computer Science, Electrical and Computer Engineering, Data Science, Artificial Intelligence, Robotics, or a closely related discipline. •        Demonstrable, hands-onexperience training or fine-tuning at least one Vision-Language-Action model(e.g., OpenVLA, Octo, Pi-0, or a comparable architecture) on a real-worldmanipulation dataset such as Bridge Data V2, Open X-Embodiment, or anequivalent collection. •        Strong proficiency in Python and PyTorch, with practical experience using the Hugging Face Transformers library and multi-GPU training workflows, including mixed-precision regimes. •        Working knowledge of ROS 2 and direct experience operating at least one physical robotic manipulator. •        Familiarity with modern vision and vision-language backbones (e.g., DINOv2, SigLIP, CLIP) and large language model components (e.g., the LLaMA family). •        Strong written and verbal communication skills, with the ability to document experimental methodology and results clearly. Preferred Qualifications•        Peer-reviewed publications invision-language modeling, robot learning, or related areas of multimodal machine learning. •        Experience with self-supervised learning or foundation-model pre-training on large-scale datasets. •        Demonstrated experience optimizing deep learning models for edge deployment, including the use of ONNX, TensorRT, and NVIDIA Jetson platforms. •        Experience leading or coordinating data-collection efforts at scale, including the supervision of teleoperators or multi-site collection operations. •        Prior contributions to computer vision pipelines for detection, tracking, optical character recognition, or related auxiliary labelling tasks. •        Recognition through academic awards, research fellowships, or competitive robotics or AI competitions. FURTHER INFORMATION & CONTACT Gross Monthly Salary Range (SGD): 5,000 to 7,000 (depending on suitability and experience). Duration of the contract: 1 year contract (renewable) Workplace address: CREATE Campus, CREATE Tower, 1 Create Way #08-01 Singapore 138602.