AI Engineer, RL

Taste Labs; San Francisco (CA); On-site; USD 175,000 - 275,000; Full time

Benefits offered by this job

Open source contributions

Experience at creative companies: Figa

Job summary

Taste Labs is building AI systems to tackle subjective “taste” problems, turning taste into measurable evaluation signals. As an AI Engineer (RL), you will help create grading methods, tasks, and environments while contributing scalable reinforcement learning infrastructure.

This onsite role in San Francisco offers a competitive USD salary and the chance to advance evaluation pipelines with leading research teams.

Qualifications

• Experience building evals, RL environments, or related ML/post-training work.
• Strong backend experience.
• Ability to translate subjective domains into verifiable systems.

Responsibilities

• Research and design grading methods and rubrics for subjective domains.
• Create tasks and environments capturing elements of ‘taste’.
• Build agent harnesses and context layers to support evaluation workflows.
• Contribute scalable RL infrastructure for training pipelines.
• Collaborate with internal research teams on training pipelines.
• Work with frontier labs on environment design to improve models.

Skills

Evals / RL environments

Backend development

Ambiguous problem solving

Tools

Python

Job description

Taste Labs is building AI systems to tackle subjective "taste" problems, where success depends on more than conventional labels. In this AI Engineer (RL) role, you will help turn taste into measurable evaluation signals by creating grading methods, tasks, and environments, while also contributing scalable reinforcement learning infrastructure. The work is designed to operate without humans in the loop, supporting reliable training and iteration at scale.

Location: San Francisco, CA (onsite). Compensation: USD 175,000 - 275,000 per year.

What you'll do

• Research and compare grading methods and rubrics for subjective domains
• Design unique tasks and define capabilities that can capture elements of "taste"
• Build agent harnesses and context layers to support evaluation and training workflows
• Contribute to scalable RL infrastructure that supports training pipelines
• Partner with internal research teams on training pipelines
• Collaborate with top frontier labs on environment design to improve frontier models

What you bring

• Experience building evals, RL environments, or related work in ML or post-training
• Strong backend experience
• Comfort working on ambiguous, hard, creative problems and translating subjective domains into verifiable systems
• A team-oriented mindset with a startup approach: move fast, adapt, take ownership, and treat scope as something to figure out rather than limit

Bonus points

• Open source contributions or personal projects that demonstrate building from curiosity
• Background at creative companies such as Figma, Notion, Canva, Adobe, or Runway
• Experience with companies focused on index building and crawling, such as Firecrawl, Brave, Luma, or Pika
• Experience in data-focused environments such as Mercor or Surge