Member of Technical Staff - Post-Training

📋 Description

• • At ReflectionAI Inc., we are on a monumental mission: to build open superintelligence and ensure its accessibility for everyone. We are pioneering the development of open-weight models designed to empower individuals, sophisticated agents, global enterprises, and even nation-states. Our distinguished team comprises leading AI researchers and seasoned company builders, many of whom have honed their expertise at prestigious institutions like DeepMind, OpenAI, Google Brain, Meta, Character.AI, and Anthropic.
• • As a Member of Technical Staff specializing in Post-Training, you will be at the forefront of transforming powerful pre-trained models into aligned, general-purpose AI agents. This role is pivotal in driving forward the research and engineering initiatives that define the cutting edge of post-training methodologies. Your work will span the entire spectrum, from meticulous data curation and sophisticated reward modeling to the implementation of large-scale optimization techniques and advanced reinforcement learning algorithms.
• • You will be instrumental in developing robust data generation pipelines, crafting effective reward models that guide AI behavior, and designing innovative reinforcement learning algorithms. Furthermore, you will explore and implement inference-time scaling techniques to ensure our models can operate efficiently and effectively at scale. This position offers a unique opportunity to contribute directly to the advancement of AI capabilities, pushing the boundaries of what large language models can achieve.
• • Collaboration is key to our success. You will work closely with our pre-training teams, fostering a seamless integration of efforts to deliver significant, step-function improvements in model capability. This cross-functional synergy will be essential in achieving our ambitious goals and ensuring our models are not only powerful but also safe and aligned with human values.
• • A significant aspect of this role involves contributing to our collective understanding of how large models learn. You will delve into the intricacies of how these models develop reasoning abilities, how they learn to follow complex instructions, and how they continuously improve through the application of reinforcement learning. Your insights and discoveries will shape our foundational knowledge in this rapidly evolving field.
• • We are seeking individuals who possess a deep and intuitive understanding of machine learning fundamentals. Practical, hands-on experience with the training of large-scale Large Language Models (LLMs) is essential. Your engineering prowess should be substantial, demonstrating a comfort level with navigating and contributing to complex ML codebases and distributed systems.
• • You should have demonstrable experience in enhancing model behavior through innovative approaches such as advanced data strategies, sophisticated reward modeling, or cutting-edge RL techniques. We look for evidence of you having independently owned ambitious research or engineering agendas that have demonstrably led to measurable improvements in model performance and alignment.
• • The ideal candidate thrives in the dynamic, high-agency environment of a startup. A strong bias toward action, coupled with clarity of execution, is paramount. You must be capable of working fluidly across both research and infrastructure boundaries, understanding the interplay between theoretical advancements and practical implementation.
• • Excellent communication capabilities are vital, as is a comfort level working collaboratively within a diverse and talented team. Your passion for advancing the frontier of artificial intelligence should be evident, driving your commitment to our mission.
• • Joining ReflectionAI means being part of building something foundational. You will help shape the future trajectory of our company and contribute to defining the very frontier of open foundational models. We are committed to enabling you to achieve the most impactful work of your career, ensuring that you and your loved ones are well-supported throughout this journey.