Member of Technical Staff (Perplexity Frontier Research)

Welcome to the application portal for Perplexity Frontier Research . This initiative develops fundamentally new algorithms, models, and evaluations. We are building an automated scientist that can conduct deep research, and provide the best quality service at the lowest cost. We are hiring research scientists for permanent, full-time positions at the initiative. One application covers all focus areas. Please refer to the Perplexity Frontier Research homepage for full details on the initiative and application. When you're ready to apply, please complete and submit the application form. Please carefully follow all instructions. Our focus areas include orchestration, pre-training, continual learning, reinforcement learning, systems, and evaluations. We will be releasing our work as papers, blog posts, and code. We'll evaluate applications on a rolling basis, with priority consideration for submissions received by November 30, 2026. You should write your application without assistance from other humans or AI models. Submitting writing from other humans or AI models as your own will result in permanent disqualification from employment at Perplexity. Qualifications: Strong foundations and engineering fluency: probability, optimization, linear algebra and algorithms, plus Python and a deep learning framework, with code that scales. Range: researchers who are proficient engineers, and engineers who are also researchers. Everyone here builds. Evidence of excellence: a result, a system, a proof, or a model that other people use or build on. A preference for understanding, written plainly: controls, ablations, honest baselines, and simple explanations. Agency, creativity, low ego: fix what's broken without being asked, ask questions that are not yet fashionable, and let ideas win on evidence. Helpful, not required: a PhD in a technical discipline, machine learning papers of outstanding quality, prior work in one of our focus areas, or large-scale empirical work with large models.