AI Research Scientist - LLMs, Agents & Scalable Systems

Perplexity; New York (NY); Sur place; USD 220 000 - 485 000; Plein temps

Résumé du poste

Perplexity seeks top-tier AI Research Scientists and Engineers to advance our Sonar models, Deep Research Agent, Comet Agent, and Search products. We aim to create SOTA experiences handling hundreds of millions of queries and scale rapidly.

You will join one of three teams—Core Research for foundational models, Agent Products for Deep Research/Canvas, or Comet Agent—driving model training, RL, and seamless integration into Perplexity’s product suite at scale.

Qualifications

• Proven experience with large-scale LLMs and deep learning systems.
• Strong programming skills in Python/PyTorch.
• Experience with post-training techniques and reinforcement learning.
• 2–6 years of relevant project experience, depending on seniority.

Responsabilités

• Research & Development: Post-train SOTA LLMs using supervised and RL techniques (SFT/DPO/GRPO).
• Leverage dataset to scale model performance across Sonar, Deep Research, Comet, and Search products.
• Stay current with LLM research, focusing on training, optimization, and personalization techniques.
• Implement preference optimization and personalization to enhance user experience.
• Invent in-house improvements to advance SOTA models.
• Turn research ideas into algorithms and run experiments to launch new models.

Connaissances

Large-scale LLMs

Python/PyTorch

Post-training techniques

Reinforcement learning

Ownership

Problem solving

Self-starter

2–6 years experience

Formation

PhD in ML/AI

Outils

Megatron

PyTorch

CUDA

Description du poste

Perplexity seeks top-tier AI Research Scientists and Engineers to advance our Sonar models, Deep Research Agent, Comet Agent, and Search products. We aim to create SOTA experiences handling hundreds of millions of queries and scale rapidly.

You will join one of three teams—Core Research for foundational models, Agent Products for Deep Research/Canvas, or Comet Agent—driving model training, RL, and seamless integration into Perplexity’s product suite at scale.