the data and evaluation behind better models
yuru is a small, senior team that builds the training data, evaluation rubrics, and eval harnesses AI labs use to train and grade frontier models, for coding models and expert domains alike. our data and rubrics support teams at OpenAI and Microsoft. we are not a body shop: we design the rubric, run the pipeline, and stand behind every score.
we also build and run our own autonomous AI in production. that is where our methods come from, and why the work we sell is proven before it reaches you, not theorized on a slide.
what we do
training data
expert-authored, task-specific datasets for coding models and specialist domains. cleaned, safety-filtered, rubric-scored, and delivered fine-tune-ready with tool-calling schemas.
evaluation rubrics
multi-dimension weighted rubrics and LLM-as-judge harnesses with chain-of-thought grading and explicit auto-fail criteria, including expert document rubrics that measure quality where benchmarks fall silent.
applied AI engineering
agentic systems and LLM apps taken to production, from browser-driving automation to real-time voice. we build them, run them, and keep them alive.
fine-tuning & consulting
post-training, model selection, and eval design, with honest answers about what AI can and can't do yet for your problem.
how we grade data
a pipeline, not a purchase
raw examples in; a quality-ranked training set out. we normalize to a fine-tune-ready schema, clean and safety-filter, score every record on a weighted rubric with an LLM-as-judge that reasons before it scores, then rank by the full score distribution and keep the top band, with per-record scores attached.
we build it too
burnt melba
a fully autonomous AI vtuber that live-streams unsupervised, drives real apps by browser automation, remembers, and sets its own goals. our proving ground for agent reliability, and where the data pipeline above was forged.
yume
our native desktop app for agentic coding: parallel agents, deep shortcuts, multi-provider. a product we ship and support.
have data, eval, or model work?
tell us what you're training or grading. we'll tell you straight whether we're the right team for it.
get in touch