Job Title: AI Research Engineer - Vision-Language Action Models Griffin Labs is looking for a Founding AI Research Engineer to lead the adaptation and deployment of Vision-Language-Action (VLA) models. We are focused on solving high-dexterity manipulation through cross-embodiment learning—training models that generalize across diverse robotic platforms and hardware configurations. The Role You will own the full development lifecycle of autonomous capabilities. This is an end-to- end mission: from reviewing raw trajectory data and designing model architectures to deployment and fleet-level monitoring. You aren’t just training models; you are building the systems that allow them to function in the real world. Key Responsibilities: End-to-End Ownership: Lead the "data-to-deployment" loop—curating datasets, fine-tuning VLA models, and building the monitoring systems that ensure fleet-level reliability. Cross-Embodiment Learning: Develop and optimize models capable of handling diverse robot morphologies and kinematic constraints. Model Adaptation: Architect and fine-tune VLA stacks that bridge high-level reasoning with low-level whole-body control. Academic Collaboration: Work alongside expert research advisors from institutions like Stanford, NTU, and SUTD to push the boundaries of what's possible in robotics. Validation: Design robust evaluation metrics in Isaac Sim/MuJoCo and bridge the Sim2Real gap for a production fleet. Requirements Technical Depth: A strong fundamental understanding of Transformers, Diffusion architectures, and LLMs/VLMs, particularly as they apply to sequential decision-making or action prediction. ML Expertise: Deep proficiency in PyTorch and experience fine-tuning large-scale multimodal models. Cross-Embodiment: A strong interest or background in generalizing AI across different robot types. Simulation: Fluency in environments like Isaac Sim or MuJoCo and the nuances of the "Reality Gap." Software Foundations: High proficiency in Python. Foundi