Vision AI Engineer for Multimodal LLMs (Equity Eligible)

Pinterest; San Francisco (CA); Hybrid; USD 139,000 - 286,000; Full time

Benefits offered by this job

Equity

Job summary

Pinterest Labs is seeking researchers to advance vision-centric large language models. You will help build VLMs that perceive visual details and aesthetics, enabling communication through visual assets with multimodal search and text-to-image tools. The role supports collaboration across a core visual pod and production-ready research.

You will contribute to evaluation benchmarks, work with Pinterest Canvas data for RLHF and fine-tuning, and publish findings to the broader research community.

Qualifications

• Experience with generative computer vision models.
• 2+ years of industry computer vision experience.
• MS/PhD in Machine Learning, Computer Science, or related areas.

Responsibilities

• Prototype new model architectures for Pinterest VLMs and finetuning.
• Develop evaluation benchmarks for vision-centric capabilities.
• Read research papers and brainstorm our visual generative strategy.
• Collect relevant visual training data for Pinterest Canvas (RLHF, fine-tuning).
• Publish and publicize your work via conferences, papers, blogs.

Skills

Generative CV models

Vision encoders

LLMs

Python / ML tooling

Education

MS/PhD in ML/CS

Tools

Cursor

Copilot

Codex

Job description

Pinterest Labs is seeking researchers to advance vision-centric large language models. You will help build VLMs that perceive visual details and aesthetics, enabling communication through visual assets with multimodal search and text-to-image tools. The role supports collaboration across a core visual pod and production-ready research.

You will contribute to evaluation benchmarks, work with Pinterest Canvas data for RLHF and fine-tuning, and publish findings to the broader research community.