AI Research/Machine Learning Engineer - Agent Products

Details

Perplexity is seeking experienced AI Research/Machine Learning Engineers and Scientists to own our Agent Products (Answering Agent, Deep Research, the Comet Browser Agent). Applicants should be well versed in all aspects of LLM prompting and training, code execution, and tool-calling in order to keep redefining what is possible with advanced retrieval.

Responsibilities

• Design a production and training system that allows for an autonomous LLM-powered agent to perform various tools across web, browsing, retrieval, and code execution in order to answer a user’s open-ended prompt/query and return a high quality answer with citations.
• Partner with the Core Research & Post-Training Team to refine the expansion, reformulation, querying, and summarization of models against Perplexity’s retrieval index.
• Devise a well-distributed set of multi-turn evaluations that help hill-climb and capture regressions for browser and deep research agents.
• Improve speed and quality of codebase and engineering processes to consistently deliver an amazing product experience.
• Collaborate with Perplexity GTM leaders and a wide breadth of stakeholders (e.g., Motorola, TripAdvisor) to develop integrated product solutions spanning research verticals like Travel and Retail.

Requirements

• Proven research experience with large-scale LLMs and Deep Learning systems.
• Strong post-training skills, familiar with supervised and reinforcement learning techniques including reward formulation and evaluation.
• Strong consumer product intuition for Agent products, willingness to unblock oneself at any level of the stack and take ownership of the end-product experience.
• Understand tradeoffs in developing an end-to-end agent including training technique, model architecture, and prompting.
• Minimum of 5 years of working on relevant deep learning research areas.

Benefits

• Cash compensation range of $200,000 to $280,000 per year.
• Equity may be part of the total compensation package.
• Comprehensive health, dental, and vision insurance for employee and dependents.
• Includes a 401(k) plan.

Perplexity has experienced tremendous growth and adoption since publicly launching the world's first fully functional conversational answer engine in 2022. Daily queries increased from 2.5 million at the start of 2024 to around 20 million by December 2024. Customers include Nvidia, Cleveland Cavaliers, Bridgewater, and Zoom. The company has raised significant funding from major technology investors and grew its employee base nearly 300% in 2024.