Research Engineer Intern (Ethical Creative and Protective AI Lab)

sonyglobal; United States; Remote; USD 57,000 - 81,000; Full time

Benefits offered by this job

Distributed team

Overtime eligibility

Job summary

SonyGlobal is offering a three-month internship focused on responsible multimodal generative AI. You will help improve model post-training and evaluation while contributing to methods and tools that make AI systems more trustworthy and reduce risks in their use.

You will collaborate with researchers and engineers, fine-tune models using supervised and RL methods, and design metrics to compare approaches for research outputs submitted to conferences or journals.

Qualifications

• PhD study in computer science, machine learning or related field (completed or current).
• Strong grounding in ML fundamentals and research methods.
• Experience with language/vision-language model tools, frameworks, APIs and prompts.
• Interest in responsible AI development and analytical problem solving.
• Excellent written and spoken English; ability to work independently.

Responsibilities

• Build post-training and evaluation infrastructure for a multimodal AI system.
• Fine-tune models using supervised and reinforcement learning methods.
• Design metrics to compare fine-tuning approaches.
• Develop tools for assessing and reducing generative AI risks.
• Collaborate with researchers and engineers and publish research outputs.

Skills

Machine learning fundamentals

Research methods

English communication

Education

PhD in CS/ML

Tools

Language model tooling

Vision-language frameworks

API integration

Prompt engineering

Job description

Role overview

Join a research and engineering team for a three-month internship focused on responsible multimodal generative AI. You’ll help improve model post-training and evaluation, while contributing to methods and tools that make AI systems more trustworthy and help address risks in their use.

Responsibilities

• Help build post-training and evaluation infrastructure for a multimodal AI system.


• Fine-tune models using supervised and reinforcement learning methods.


• Design and implement metrics to compare fine-tuning approaches.


• Develop tools and frameworks for assessing and reducing generative AI risks.


• Collaborate with researchers and engineers and contribute to research outputs for conferences or journals.

Requirements

• Completed or current PhD study in computer science, machine learning, or a related scientific field.


• Strong grounding in machine learning fundamentals and research methods.


• Experience with language or vision-language model tools, frameworks, API integration, and prompt engineering.


• Interest in responsible AI development and strong analytical and problem-solving skills.


• Clear written and spoken English communication; ability to work independently and propose ideas.

Nice to have

• Hands-on experience with reinforcement learning fine-tuning of language or vision-language models.

Benefits and work setup

• Three-month internship with a geographically distributed team.


• Target pay is $50 per hour, with overtime eligibility.