AI Engineer & Researcher – Post-training – HYBRID

Job Overview

The post-training team at xAI transforms powerful pre-trained models to become steerable, versatile, and capable of understanding and addressing real-world challenges. As a post-training researcher/engineer, you will enhance the model’s instruction-following capability and general usefulness to fulfill our mission – developing AI systems that can accurately understand the universe, create new knowledge, and improve themselves through interactions. Responsibilities include creating and driving research agenda to advance model quality, improving data mixtures by building data collection pipelines and developing data generation techniques, creating generalizable reward models and developing novel reinforcement learning algorithms, designing and implementing robust model evaluations, designing and implementing large-scale model training frameworks, and collaborating with pre-training, reasoning, data, multimodal, applied, product efforts to push the frontiers of model capability. Ideal candidates are experts in ML and fine-tuning large language models, have a track record in leading research that significantly impacts AI advancement, experience in data-driven large language model behavior improvements, advanced reinforcement learning or inference-time search techniques, developing benchmarks or large-scale distributed machine learning systems, and model optimizations under complex setups. The role is based in the Bay Area (San Francisco and Palo Alto), and candidates are expected to be located near the Bay Area or open to relocation. The tech stack includes Python, Jax, and Rust. The interview process includes an initial phone interview, followed by four technical interviews including coding assessment, post-training technical sessions, and a presentation of past exceptional work and vision with xAI. All interviews are conducted via Google Meet.