AI Platform Engineer

📋 Description

• • Join Fastino AI, a rapidly growing and well-funded startup at the forefront of developing the next generation of Large Language Models (LLMs). Our mission is to create specialized, efficient AI, building upon the success of our GLiNER family of open-source models, which have garnered over 5 million downloads and are utilized by industry giants like NVIDIA, Meta, and Airbnb. With a recent $25M seed funding round backed by prominent investors including Microsoft, Khosla Ventures, Insight Partners, and key tech leaders, we are poised for significant expansion and innovation.
• • As an AI Platform Engineer, you will play a pivotal role in owning and evolving Fastino's end-to-end model platform. This is a foundational, systems-level engineering position, distinct from feature development, where you will be instrumental in shaping the infrastructure that powers our AI advancements. Your responsibilities will span the entire lifecycle of model development and deployment, ensuring robustness, scalability, and efficiency.
• • You will be responsible for designing and building robust training pipelines, enabling the efficient development of our cutting-edge AI models. This includes architecting distributed fine-tuning workflows tailored for both small encoder and decoder models, incorporating advanced techniques such as LoRA, adapters, distillation, and compression to maximize performance and resource utilization.
• • A key aspect of your role will involve establishing comprehensive experiment tracking, ensuring reproducibility, and implementing sophisticated dataset versioning systems. This will provide a solid foundation for iterative development and rigorous scientific inquiry, allowing us to manage and understand the vast complexity of our AI research.
• • You will focus on optimizing training efficiency across multiple dimensions, including GPU utilization, memory management, throughput, and cost-effectiveness. This optimization is critical for maintaining a competitive edge and enabling rapid iteration in a resource-intensive field.
• • Furthermore, you will design and implement scalable Reinforcement Learning (RL) training workflows, encompassing policy optimization and reward modeling. This includes integrating RL methodologies seamlessly with supervised fine-tuning and distillation processes, unlocking new capabilities and performance enhancements for our models.
• • Building robust evaluation loops and automated regression detection mechanisms will be a core responsibility, ensuring the quality and reliability of our models throughout the development cycle. This proactive approach to quality assurance is paramount for production-ready AI systems.
• • You will architect and build scalable data ingestion pipelines, capable of handling both structured and unstructured data sources. This involves designing efficient data curation, filtering, and quality enforcement systems to ensure the integrity and utility of the datasets used for training.
• • Implementing reproducible data workflows that are tightly coupled with training runs will be essential for maintaining consistency and enabling seamless collaboration within the team.
• • On the deployment front, you will architect low-latency inference services, ensuring our models can be served efficiently and effectively in production environments. This includes designing secure and reliable production deployment workflows, adhering to best practices for operational stability and performance.
• • This role offers a unique opportunity to shape the core infrastructure of a leading AI research company, working with a world-class team and leveraging state-of-the-art technologies. You will have a direct impact on the development and deployment of advanced AI models that are already making waves in the industry.