Skip to content
Project Summary
Welcome to Project Selenite! Here's everything you need to know to complete this project successfully.
Last edited 2 days ago by Kelly Victoria.
🚨 This project is strictly confidential. Please don't share with anyone outside our project team.
Project Workflow
Let’s dive into the workflow at a high level. This is a complex project, and it can be a steep learning curve if this is your first LLM training project with our team at Pareto! We hope you’ll stick with us, as this gets much easier with practice! We’re also hoping to work with you across many more projects if this works out!
Feel free to ask any questions if you’re unsure about anything.
The requester is building a system that evaluates AI agents by having them interact with simulated websites (“gyms”). These gyms mirror real-world tools like Redfin, Airbnb, or Linear, and are used to test how well an AI agent can complete realistic user tasks, such as browsing properties, managing bookings, or tracking projects.
Your goal is to extract structured context from these gyms, including tool overviews, environment setups, capabilities, limitations, atomic actions, and UI mappings.
Your contributions will help to:
Initial Timeline
Target: 300 Tasks for each Gym
Timeline:
💰Compensation
Point of Contact
🤝 Point of Contact
Data Delivery Lead:
@Mahima Joshi
Project Manager:
@Kelly Victoria
Support Team
Expectations
As part of this project team, you’re responsible for ensuring we deliver the highest quality work to our researchers. We look for your insights, feedback, and commitment to quality work to ensure we can keep this requester and their projects around for you long term.
We risk losing future projects for this entire team when we deliver recurring poor quality results to the researcher.
Once you’re approved for this project’s main batch, you’re officially part of the requester’s team and have priority access to all future projects by this researcher. To remain in good standing, please prioritize the quality of your work and work with your reviewers to integrate feedback. We will prioritize those with the highest quality and consistency work to take on greater roles in existing and future projects.
Quality > Quantity
We’re here to support the learning curve!
This is a complex project, and it can be a steep learning curve if this is your first LLM training project with our team at Pareto! We hope you’ll stick with us, as this gets much easier with practice! We’re also hoping to work with you across many more projects if this works out!
Feel free to ask any questions if you’re unsure about anything.
Confidentiality Reminder
In an effort to protect the quality, integrity, and diversity of the models we are training, the submission and use of AI generated content while working on projects is strictly forbidden unless otherwise permitted by guidelines. Read more about the dangers of and .
Refrain from using any type of AI when working on Pareto projects.
Any type of AI use is prohibited when working on Pareto projects, unless explicitly allowed by the requester. This includes supplementary items like AI-translators, AI-thesaurus addons, Grammarly predictive-text, generative-fill, and directly using LLMs to create any type of content.
Prompts created with these types of AI tools are not usable by the requester, and can lead to payout delays or removal from the project.
In addition, plagiarizing and using content from third party sources is prohibited because it puts Pareto at risk for IP infringement/theft lawsuits. Each prompt should be created uniquely for each separate project. Do not use existing datasets, blocks of questions online, copying of textbook question sets or quizzes, no academic or private test questions (like Mensa), and do not reuse prompts from previous projects.
We appreciate your cooperation on this!
Next, let’s get you started with
Want to print your doc?
This is not the way.
This is not the way.
Try clicking the ⋯ next to your doc name or using a keyboard shortcut (
CtrlP
) instead.