Research Scientist (US)

📋 Description

• • Join Fastino AI as a Research Scientist and be at the forefront of developing the next generation of specialized and efficient Large Language Models (LLMs). Our ambitious mission is to push the boundaries of AI, creating models that are not only powerful but also highly optimized for specific tasks and domains. You will be an integral part of a world-class team, comprised of alumni from prestigious institutions and leading tech companies such as Google Research, Apple, Stanford, and Cambridge. Our commitment to open-source innovation is evident in our GLiNER family of models, which have achieved over 5 million downloads and are actively utilized by industry giants like NVIDIA, Meta, and Airbnb. Fastino has secured significant funding, raising $25 million in our seed round, as highlighted in TechCrunch, and is backed by a distinguished group of investors including Microsoft, Khosla Ventures, Insight Partners, and prominent tech leaders like Github CEO Thomas Dohmke and Docker CEO Scott Johnston.
• • In this role, you will have the opportunity to directly influence Fastino's research trajectory. Your primary responsibilities will involve conducting cutting-edge research and experimentation with novel language model architectures. You will play a crucial role in driving and executing Fastino's research roadmap, contributing to the development of groundbreaking AI technologies. A key focus will be on optimizing our multimodal models to enhance response quality, improve instruction adherence, and elevate overall performance metrics. This involves a deep dive into model behavior, identifying areas for improvement, and implementing sophisticated solutions.
• • You will also be instrumental in architecting and refining our data processing pipelines. This includes implementing advanced filtering, balancing, and captioning systems to guarantee the highest quality of training data across a diverse range of content categories. Ensuring data integrity and diversity is paramount to building robust and unbiased AI models. Furthermore, you will implement state-of-the-art reinforcement learning techniques, such as Direct Preference Optimization (DPO) and Generalized Reward Preference Optimization (GRPO), to align model outputs precisely with human preferences and desired quality standards. This hands-on application of advanced RL methods will be critical in shaping the user experience and reliability of our models.
• • Building robust, real-world motivated evaluations will be another core aspect of your work. You will design and implement comprehensive evaluation frameworks that accurately reflect the challenges and nuances of deploying AI in practical scenarios. This includes developing metrics and methodologies to rigorously assess model performance, safety, and fairness. Close collaboration with the Fastino engineering team is essential, as you will work together to ship model updates directly to our customers, ensuring a seamless transition from research to production. Your contributions will directly impact the products and services our users rely on.
• • Finally, you will play a vital role in establishing best practices for code health and documentation within the research team. This proactive approach will foster a collaborative environment, enhance code maintainability, and ensure reliable development cycles, making our research efforts more scalable and sustainable. This position offers a unique chance to make a significant impact in the rapidly evolving field of AI, working alongside a passionate and talented team dedicated to building the future of intelligent systems.