Data Scientist at HERE Technologies in Chicago, IL

Summary: HERE Technologies is a location data and technology platform company. They are seeking a Data Scientist II to develop and scale large-scale visual models, specializing in computer vision and self-supervised learning. Responsibilities: - Model Development: Design, train, and scale vision foundation models across image and video modalities - Fine-tuning & Adaptation: Utilize transfer learning, fine-tuning, and adapter methods to adapt models to specific downstream tasks - Pipeline Engineering: Create efficient data pipelines and pre-training strategies for large-scale datasets - Optimization: Optimize model architecture, training efficiency, latency, and throughput - Research to Production: Implement novel algorithms from research papers to solve real-world computer vision challenges - Evaluation: Implement evaluation frameworks (e.g., using LLM-as-a-judge or traditional metrics) to measure model performance and quality Required Qualifications: - 1-2 years of experience, specializing in Vision Foundation Models - Develops and scales large-scale visual, multimodal, or generative models (like ViT, SAM, DinoV3) using Python and PyTorch/TensorFlow - Experience in self-supervised learning, computer vision, and deploying high-performance models - Design, train, and scale vision foundation models across image and video modalities - Utilize transfer learning, fine-tuning, and adapter methods to adapt models to specific downstream tasks - Create efficient data pipelines and pre-training strategies for large-scale datasets - Optimize model architecture, training efficiency, latency, and throughput - Implement novel algorithms from research papers to solve real-world computer vision challenges - Implement evaluation frameworks (e.g., using LLM-as-a-judge or traditional metrics) to measure model performance and quality - Advanced proficiency in Python and deep learning frameworks like PyTorch - Strong understanding of Computer Vision techniques, Vision Transformers (ViT), and self-supervised learning - Hands-on experience with foundational model architectures (e.g., SAM, DinoV3) - Experience handling large-scale, unstructured datasets - Typically requires a Master's or above in Computer Science, AI, or related fields - Experience with CUDA for performance optimization Preferred Qualifications: - Experience with generative models or diffusion models - Knowledge of Multimodal Learning (image and text/video) - Experience with edge AI deployment - Experience with database management and optimization, with a strong preference on PostgreSQL / PostGIS Required Skills: Python, PyTorch, TensorFlow, Computer Vision, Vision Transformers, Self-supervised Learning, Vision Foundation Models, Model Architecture Optimization, CUDA, Large-scale Data Handling Benefits: Health (Medical/Dental/Vision) insurance, Retirement savings plans, Paid time off & leave policies