A Data Labeling Platform that enables users to generate valuable datasets for model training and fine-tuning
#Sahara AI #Data Annotation #English
Scope
- Build a task marketplace where individual labelers can find decent labeling tasks
- Provide tools for annotating and reviewing datapoint
User
I have a background in Web3 and discovered Sahara’s Data Service Platform through media exposure and campaign activities. Sahara plans to airdrop rewards to contributors in the future. To qualify for the airdrop, I aim to earn as many Sahara Points as possible by completing labeling tasks on the platform.
Basic Info
My Role
Product Designer
Date
Aug, 2024 - May 2025
Data by Oct, 2025
170K+approved datapoints
Task Flow
Use Cases
Select a task: Browse through different tasks and pick the ones that you find interesting and suit your skills.
Task list
Task detail page
Demo
Qualify: Demonstrate your expertise and qualifications for the task by completing a short exam.
Take an exam
Exam tool
Demo
Contribute: Complete annotating or reviewing tasks based on assigned role
Labeling Tool for annotator
Pinned datapoints
Reviewing tool for reviewer
Demo
Reward: Receive points and rewards for each task you complete. Get bonus rewards for accuracy and high rankings.
Demo
Design Research
Competitors
Findings
Design Explorations
1. Explore the Structure of a Task Card
For quick demo, we used the Ant Design System to build our product in the MVP version.
2: Labeling Tool Drafts
Focus on finding a decent way for switching dps and submission.
Draft A
Draft B
Draft C
When can a labeler submit a batch of data points?
Complete all required questions for each data point in a batch. Then submit the entire batch.
Draft A, Draft B → Draft C
Draft A: Allows labelers to easily move to the next data point after completing one.
Draft B: Allows labelers to switch back and forth within a batch of data points.
However, Draft A and B do not clearly connect the relationship between switching and submission. Draft C addresses this gap by combining the strengths of both Draft A and Draft B, creating a more seamless workflow.
