AI Engineer, RL & Evals

Taste Labs is building the data and infrastructure layer for taste. The goal is to end AI slop: to make AI feel right, not just be correct. The company raised $18.5M in seed funding co-led by Amplify and CRV, and most frontier labs are already customers.

AI has nailed objective domains and can generate anything. The hard part left is judgment: what fits, what feels like you, what is actually great. Taste Labs is turning that into something measurable, starting with design. They do it on two sides: building the post-training data and RL environments that teach taste to frontier models, and the context and verification tools agents need to produce work that is more creative, more on-brand, and more right.

THE OPPORTUNITY

How do you judge something non-verifiable, like design? To solve subjective domains, you have to solve how to grade the difference between slop and great without the need for humans in the loop. As AI Engineer, RL & Evals, you will work at the core of that problem, building the evaluation frameworks and reinforcement learning infrastructure that make subjective quality measurable at scale.

WHAT YOU’LL DO

• Research grading methods and rubrics for evaluating subjective, non-verifiable outputs like design quality


• Design tasks that can capture elements of taste and design capabilities at scale


• Build agent harnesses and context layers that improve model output quality


• Work on scalable RL infrastructure supporting post-training pipelines


• Collaborate with internal research teams on training pipelines


• Partner with top frontier labs to craft environments that improve frontier model capabilities

YOU SHOULD HAVE

• Experience building evals, RL environments, ML systems, or post-training pipelines, with strong backend engineering foundations


• Comfort with ambiguous, hard, creative problems and a drive to make subjective domains verifiable


• Startup DNA: you move fast, adapt, take ownership, and treat nothing as out of scope

NICE TO HAVE

• Open source contributions or personal projects that demonstrate you build things out of curiosity


• Background at creative companies such as Figma, Notion, Canva, Adobe, Runway, or Pika


• Experience at companies with strong index building or crawling, such as Firecrawl, Brave, or Luma


• Background in data infrastructure at companies like Mercor or Surge