MLE L5 – AR Veritas

AR Veritas is seeking a Machine Learning Engineer to build systems for provably correct AI reasoning with LLMs. The role involves creating automated data annotation pipelines, fine-tuning LLMs, and developing experimental test harnesses. This is a 12-month extendable contract, based remotely in Berlin, collaborating with teams in Berlin, Boston, and Seattle.

• Work directly with science and engineering teams to build systems for provably correct AI reasoning with LLMs
• Build a system that generates automatically annotated data using an existing reasoning engine (AI planner, simulator, proof planner)
• Handle large-scale data annotations and automate pre/post-processing pipelines, including data lake management in S3 and compute resources
• Build pipelines for training and fine-tuning LLMs with data
• Build an experimental test harness able to run experiments in multiple domains with minimal adaptation
• Take part in scrum development in an international setting

• 3+ years of experience with training/deploying deep learning/ML models
• Strong knowledge of relevant ML tools (PyTorch, Hugging Face, SageMaker, vLLM)
• Experience in scripting for automation (e.g., Python)
• Experience using cloud storage and computing technologies (AWS, EC2, S3, SageMaker, Lambda)