AI Research Scientist - Post-Training LLMs & Agent Systems

Perplexity; San Francisco (CA); On-site; USD 220,000 - 485,000; Full time

Job summary

Perplexity is seeking top-tier AI Research Scientists and Engineers to advance our AI products and capabilities, including Sonar models, Deep Research Agent, Comet Agent, and Search. You will work on one of three specialized teams and own key research and development tasks—from post-training LLMs to optimization and personalization—driving next-generation experiences.

Collaborate with engineering to deploy training pipelines on Megatron/PyTorch, build scalable infrastructure, and turn ideas into

Qualifications

• Proven experience with large-scale LLMs and deep learning systems.
• Strong Python and PyTorch programming skills.
• Experience with post-training techniques and reinforcement learning.
• Self-starter with ownership of tasks.
• Passion for tackling challenging problems.
• 2–6 years of experience on relevant projects.

Responsibilities

• Develop post-train LLMs using SFT/DPO/GRPO.
• Scale model performance across Sonar, Deep Research, Comet, and Search products.
• Stay current with the latest LLM research, especially in model training, optimization, and personalization techniques.
• Implement preference optimization and personalization capabilities to enhance user experience.
• Invent in-house improvements and optimizations to enhance SOTA models.
• Turn research ideas into algorithms and run experiments to launch new models.

Skills

LLMs

Deep Learning

Python

PyTorch

Reinforcement learning

Post-training techniques

Ownership

Self-starter

Education

PhD in Machine Learning

Tools

Megatron

CUDA

Job description

Perplexity is seeking top-tier AI Research Scientists and Engineers to advance our AI products and capabilities, including Sonar models, Deep Research Agent, Comet Agent, and Search. You will work on one of three specialized teams and own key research and development tasks—from post-training LLMs to optimization and personalization—driving next-generation experiences.

Collaborate with engineering to deploy training pipelines on Megatron/PyTorch, build scalable infrastructure, and turn ideas into