Strong, polyvalent programming skills in Python covering parallel computing, system design, large-scale deployments, AWS deployments and model evaluations
Experience developing and maintaining multimodal data pipelines
Experience in training and deploying LLMs, VLMs or Pytorch models
MSc or PhD in machine learning, computer vision, natural language processing, or a related field
Deep understanding of training and evaluation paradigms for multimodal models
Responsibilities:
Develop and implement cutting-edge data strategies to improve the performance, efficiency, and applicability of LLMs, VLMs and Action Models
Generate and augment synthetic multimodal datasets, including images, text, and action trajectories, to advance model capabilities in areas like VQA, agent behaviors, and virtual navigation
Apply model distillation techniques to optimize large-scale models for edge deployment, ensuring scalability without compromising performance
Design and iterate on evaluation frameworks to target edge cases and measure model improvements across multiple domains
Lead research into aligning data with human and AI preferences, implementing feedback loops to refine agent decision-making and learning behaviors
Collaborate effectively with cross-functional teams to integrate data-driven solutions into LLM, VLM and Agent systems
Stay at the forefront of breakthroughs in AI data strategies, model distillation, and multimodal learning through active scientific exploration