RL environments
Expert-built, containerized tasks for long-horizon agent work, with verifiers, proof of solvability, and pass rates.
The Data Lab specializing in code
Train and evaluate everything frontier AI does through code
Apps, games, interfaces, infrastructure, and work beyond software in science, ML, ops, and security. If models produce it through code, it’s in our domain.
Trusted by
Data for every domain, from software to science.
We specialize in the domains where models build, run, and fix real systems through code.
Software engineering
Work across the software lifecycle, from understanding and debugging an unfamiliar codebase to building features and writing tests.
Terminal coding
Command-line work, such as configuring systems, writing scripts, running and fixing builds, and managing services and data.
Frontend and web apps
Web pages, interfaces, and full web apps, including layout, styling, interaction, and accessibility.
Infrastructure as code
Cloud infrastructure defined and changed in code, such as networks, permissions, deployment pipelines, and monitoring.
Game development
Building games and interactive worlds in code, from rules and physics to levels, puzzles, and player input.
3D graphics
Scenes, 3D models, and assets made through code and 3D tools, including geometry, materials, lighting, and animation.
Machine learning engineering
Preparing data, training models, and evaluating, tuning, and serving them in production.
Scientific computing
Science done through code, such as data analysis, simulation, numerical methods, and research software.
Data, environments, evals, and experts.
License pre-built data, co-design something new, or bring our experts onto your platform.
Data and environments
SFT data
Demonstrations authored or improved by experts, with chain-of-thought, diffs, and trajectories where you need them.
Preference data
Experts compare, rank, and rate your model’s outputs for RLHF, DPO, and reward models.
Evals
Domain experts judge outputs against task-specific rubrics. Each eval ends with a report recommending the data and training to close the gaps.
Services
Off-the-shelf data
Pre-built, expert-authored datasets and RL environments with tasks, verifiers, and trajectories, ready to license.
Bespoke data
Our applied AI research team co-designs data and environments with you for specific domains and failure surfaces.
Expert staffing
Vetted experts from our network of 400,000+ work on your platform, managed by your team or our leads.
Research
Benchmarks, leaderboards, and analyses from our research team, including the Code Index.
Don’t see what you need? Ask us
Research and datasets from our lab.
Explore ready-to-use collections, independent benchmarks, and our latest findings.
Leaderboard
Code Index
A live coding leaderboard on private collections that stay unpublished.
- 01Claude Opus 548.3
- 02GPT-6 Astra45.0
- 03Claude Fable 542.8
- 04GPT-5.6 Sol39.4
Benchmark
UI Quality Bench
10 frontier models, 35 hard prompts, senior designers ranking every result blind.
Frontier AI needs quality code. Our expert network authors it.
For more than ten years, Revelo has hired, tested, and placed senior engineers in real jobs. Today, those same engineers author the datasets, RL environments, and evals that leading AI labs use to train and evaluate their models.
Hired and tested
Our tasks are authored by engineers who passed our hiring process and hold real jobs.
Matched by specialty
Each task goes to engineers who do that kind of work in their own jobs.
Graded and reviewed
Every task is graded by tests or rubrics and reviewed by someone who does that work.
How is Revelo different from a general expert network?
General networks recruit across many fields. We focus on engineering, and each task goes to people who do that work every day.
Do you only cover software engineering?
No. We cover anything a model builds through code, including frontend and web apps, games, 3D graphics, infrastructure, command-line work, ML engineering, and scientific computing. If you need a domain that isn’t on this list, we’ll scope it with you.
Your collections share names with public benchmarks. Are they the same tasks?
No. They follow the same formats, but our engineers author new tasks, many from merged pull requests that fixed real issues.
How is task difficulty calibrated?
A solvability review first removes tasks that are broken or ambiguous. We then run frontier models against the collection before it ships and report the pass rate, with the model, the attempt count, and the grading method attached.
Who are your experts, and how are they vetted?
Engineers from our network of 400,000+. Each one passes technical assessments, manual reviews, and fraud checks, the same vetting we’ve used for more than ten years of hiring. We then match each task to engineers who do that kind of work in their own jobs.
What do I receive?
Each RL environment ships as a container with the task, a verifier, proof that it can be solved, and its pass rate. SFT data includes chain-of-thought, diffs, and trajectories where you need them.
What formats do datasets ship in?
SWE-bench style task sets, Harbor-format terminal tasks, agent trajectories, and DPO-ready preference pairs. If you need a format that isn’t on that list, we’ll match it.
Can you work inside our stack and QA process?
Yes. Our engineers can work in your tools and follow your QA steps, or deliver finished data from our platform with our own review. You can also combine the two.
Can you build custom benchmarks and evals for our use case?
Yes. We build task collections calibrated to a difficulty you set, execution-graded environments, and evals scored against your rubric.
Put our engineering
expertise to work.
A researcher will help you plan the data, environments, and evals your model needs.