Research Engineer, Gemini Multi-turn

Google; Mountain View (CA); On-site; USD 180,000 - 240,000; Full time

Benefits offered by this job

Bonus Target

Equity

Benefits

Job summary

Google is seeking researchers to advance Gemini LLMs, focusing on multi-turn capabilities, factuality, and tool-use. The role involves collaborating with cross-functional teams to develop supervised fine-tuning and reinforcement learning experiments for production deployment.

Required are a CS or AI Bachelor's degree and 5 years of Python experience, plus a track record of first-author LLM publications. RLHF and distributed deep learning frameworks experience is essential.

Qualifications

• Requires a Bachelor's degree in CS or AI and 5 years of Python experience.
• Must have first-author publications in LLM conferences.
• Experience with RLHF and distributed deep learning frameworks.

Responsibilities

• Research and develop approaches to enhance Gemini LLMs' multi-turn capabilities, factuality, and tool-use.
• Collaborate with cross-functional teams to implement supervised fine-tuning and RL experiments for production deployment.

Skills

Python

LLMs

SFT

RLHF

DPO

PPO

JAX

Flax

Distributed Systems

Model Alignment

Generative AI

Multi-turn Modeling

Long-context Reasoning

Tool-use

Function Calling

Search Grounding

Education

Bachelor's degree in CS or AI

Job description

Research and develop new approaches to enhance the multi-turn capabilities, factuality, and tool-use of Gemini LLMs. Collaborate with cross-functional teams to implement supervised fine-tuning and reinforcement learning experiments for production deployment.

Requirements: Requires a Bachelor's degree in CS or AI with 5 years of Python experience and a track record of first-author publications in LLM conferences. Experience in post-training techniques like RLHF and working with distributed deep learning frameworks is essential.

Key Skills: Python, Large Language Models, SFT, RLHF, DPO, PPO, JAX, Flax, Distributed Systems, Model Alignment, Generative AI, Multi-turn Conversational Modeling, Long-context Reasoning, Tool-use, Function Calling, Search Grounding

Benefits: Bonus Target, Equity, Benefits