Machine Learning Engineer II, Computer Vision Applied Science
About this role
Pinterest is seeking a Machine Learning Engineer II to join its visual modeling team, focusing on developing vision-centric large language models. The ideal candidate will possess experience in generative computer vision and contribute to the advancement of AI/ML initiatives at Pinterest.
What you'll do
- Prototype new model architectures for vision-centric VLMs
- Develop evaluation benchmarks for visual capabilities
- Engage in research discussions and brainstorm strategies
- Assist in gathering relevant visual training data
- Publish research findings at conferences and in publications
- Mentor junior researchers and interns
What they're looking for
- Experience with generative computer vision models
- 2+ years in industry computer vision
- M.S. or PhD in Machine Learning, Computer Science, or related fields
- Familiarity with visual encoders and LLMs
- Experience with AI coding assistants
- Ability to conduct research and analysis
Benefits
- Flexible work environment
- Opportunities for mentorship
- Collaborative team culture
- Chance to publish research
- Engagement with a diverse tech community
- [unknown]
Opens the official application on the employer’s site. No login required.
Pinterest builds a large-scale platform serving millions of users, with infrastructure spanning security, database systems, and mobile products, supported by data systems and advertising technology. The company is hiring Software Engineers II and experienced engineers across security, infrastructure, iOS development, and data engineering to enhance platform capabilities, improve detection and response systems, optimize performance, and leverage AI-driven solutions.
- Website
- pinterest.com
Likely interview questions
- Describe your experience finetuning open-source LLMs for vision tasks. What challenges did you encounter and how did you address them?
- Walk us through how you've designed evaluation benchmarks for computer vision models. How would you approach creating benchmarks for vision-centric capabilities like fashion style recommendations?