Snowflake AI Research Scientist New Grad – Agents & RL | Bellevue | $176k–$230k

Snowflake; Bellevue (WA); On-site; USD 176,000 - 230,000; Full time

Job summary

Snowflake seeks a New Grad AI Research Scientist – Agents & RL in Bellevue, WA. You will tackle agentic AI frameworks, reinforcement learning, and scalable research that can move to production within the Snowflake Data Cloud.

You'll design coding and auto-research agents, run RLHF, DPO, and PPO work, and contribute to publishing in venues like NeurIPS, ICML, ICLR, or ACL.

Qualifications

• PhD in CS/ML/AI or equivalent research experience.
• Foundational RL experience required.
• Publication record in AI/ML.

Skills

RL experience

LLM research

Agentic planning

Self-correction architectures

Publications

Education

PhD in CS/ML/AI

Tools

PyTorch

JAX

Python

Job description

Snowflake AI Research Scientist New Grad – Agents & RL | Bellevue | $176k–$230k

Role AI Research Scientist, New Grad – Agents & Reinforcement Learning

Company

Company Snowflake

Location(s)

Location(s) Bellevue, WA

Term

Term Full-time new grad

Pay

Pay $176,000–$230,000

Deadline

Deadline Rolling — apply early

Overview

This Bellevue role sits where agentic AI meets reinforcement learning inside the Snowflake Data Cloud. You design frameworks that let agents refine themselves, build coding and auto-research agents, and run RLHF, DPO, and PPO work that changes how enterprises automate.

Store Big Data

The posting expects research that can also show up in production systems used at enterprise scale, with a path to publish at venues such as NeurIPS, ICML, ICLR, or ACL.

Eligibility / Key Requirements

• PhD in Computer Science , Machine Learning , Artificial Intelligence, or a closely related field (completing, recently completed, or equivalent research experience)
• Foundational RL experience (RLHF, DPO, PPO, or multi-agent systems)
• Research in LLM post-training, fine-tuning, or reasoning models
• Ability to implement agentic architectures (tool use, planning, self-correction)
• Python plus PyTorch or JAX; at least one first-author or co-authored publication or preprint in a relevant AI/ML area