Agentic Engineer - LLM and Reinforcement Learning

Company Description

YC Bench is a real-time platform designed to benchmark and evaluate forecasting models that predict which Y Combinator startups will excel beyond their peers in each batch. By combining data-driven insights and the latest in predictive analytics, YC Bench provides a unique tool for assessing startup potential. Our mission is to enhance decision-making in the startup ecosystem by leveraging advanced technology and innovative methodologies. YC Bench operates at the intersection of artificial intelligence, innovation, and entrepreneurship.

Role Description

This is a full-time, remote role for an Agentic Engineer specializing in LLM (Large Language Models) and Reinforcement Learning (RL). The role involves designing, developing, and deploying advanced AI and RL models to enhance predictive capabilities. Responsibilities include optimizing model performance, building scalable pipelines, and collaborating with cross-functional teams. You will also be tasked with conducting experiments, analyzing data, and iterating on algorithms to improve overall accuracy and efficiency.

Qualifications

• Expertise in developing, fine-tuning, and deploying Large Language Models (LLMs)
• Proficiency in Reinforcement Learning techniques, including knowledge of relevant frameworks and algorithms
• Strong Python programming skills and experience with machine learning libraries (e.g., TensorFlow, PyTorch, Hugging Face)
• Experience in data preprocessing, data pipelines, and managing large datasets
• Strong analytical and problem-solving skills with an aptitude for mathematical foundations of AI
• Experience in software engineering, including version control (Git), debugging, and performance optimization
• Excellent communication skills and ability to collaborate effectively in a remote team environment
• Advanced degree in Computer Science, Machine Learning, or a related field (Ph.D. preferred but not required)
• Experience in the startup ecosystem and an understanding of Y Combinator’s processes is a plus