Applied Research Engineer

About the Role

As an Applied Research Engineer at Labelbox, you will be at the forefront of developing cutting-edge systems and methods to create, analyze, and leverage high-quality human-in-the-loop data for frontier model developers. Your role will involve designing and implementing advanced systems that align human feedback into AI training processes, such as Reinforcement Learning from Human Feedback (RLHF), Direct Preference Optimization (DPO), etc. You will also work on innovative techniques to measure and improve human data quality, and develop AI-assisted tools to enhance the data labeling process. Your expertise in machine learning, frontier model training, and advanced human data alignment techniques will be crucial in pushing the boundaries of AI capabilities and delivering state-of-the-art solutions to meet the evolving needs of our customers.

This role is based in our dedicated tech hub in San Francisco, CA. We use a hybrid work model of 2 days in the office per week.

Your Day to Day

• Conduct cutting-edge research on advanced methods for aligning human preferences with AI systems, including RLHF and other novel approaches.

• Design and develop rigorous systems to measure, enhance, and leverage the quality of human-in-the-loop data for AI training.

• Create AI-assisted tools that incorporate active learning and adaptive sampling techniques to increase the efficiency and effectiveness of the human data labeling process.

• Investigate the impact of different types of human feedback (e.g., demonstrations, preferences, critiques) on model performance and alignment.

• Develop and implement novel algorithms for learning from human preferences and for optimizing the human feedback collection process.

• Collaborate with engineering and product teams to integrate research findings into Labelbox's product suite, focusing on scalable and practical applications of human-AI alignment techniques.

• Engage with customers and the AI community to understand evolving human data needs for frontier models and to share insights on best practices.

• Publish research findings in top-tier academic journals and present at leading AI conferences.

• Stay at the forefront of advancements in AI, particularly in areas related to human data quality, human-AI collaboration, and AI alignment.

• Contribute to technical documentation, blog posts, and educational content to establish Labelbox as a thought leader in human-centric AI development.

About You

• Ph.D. or Master's degree in Computer Science, Machine Learning, AI, or related field

• At least 3 years of experience addressing sophisticated ML problems with successful delivery to customers

• Track record of designing robust data quality measurement and refinement systems for improving model performance

• Deep understanding of frontier models (e.g., large language models, multimodal models), state-of-the-art post-training methods, and their human data requirements

• Proficiency in programming languages such as Python, and experience with deep learning frameworks (e.g., PyTorch, JAX, TensorFlow)

• Excellent research skills with a track record of publications in top-tier AI/ML venues (e.g., ACL, EMNLP, NAACL, NeurIPS, ICML, ICLR, etc.)

• Adept at interpreting research literature and quickly turning new ideas into prototypes

• Strong analytical and problem-solving abilities

• Excellent communication skills and ability to collaborate in a multidisciplinary team

Labelbox Applied Research

At Labelbox Applied Research, we're committed to pushing the boundaries of AI and data-centric machine learning, with a particular focus on advanced human-AI interaction techniques. We believe that high-quality human data and sophisticated human feedback integration methods are key to unlocking the next generation of AI capabilities. Our research team works at the intersection of machine learning, human-computer interaction, and AI ethics to develop innovative solutions that can be practically applied in real-world scenarios.

We foster an environment of intellectual curiosity, collaboration, and innovation. We encourage our researchers to explore new ideas, engage in open discussions, and contribute to the wider AI community through publications and conference presentations. Our goal is to be at the forefront of human-centric AI development, setting new standards for how AI systems learn from and interact with humans.