19-Year-Old Chinese-American Takes on Scale AI: Turning AI Data into a Gaming-Like Way to Earn
Datacurve: Turning Data Annotation into a Gamified Engineer Arena
Artificial intelligence has gone viral — and it has taken the “pick-and-shovel” data annotation business along for the ride.
- Scale AI now has a valuation above $20 billion.
- Surge AI is reportedly raising around $1 billion.
- Tech giants and investors are chasing one thing: high-quality training data.
In this frenzied environment, Serena Ge, a 19-year-old Chinese-American entrepreneur, and her team of just 10 people, have raised $15 million (about 100 million RMB) for Datacurve.
Datacurve counts Chemistry VC, Y Combinator, and engineers from DeepMind, Anthropic, and OpenAI among its investors — all drawn to a bold idea: turning high-quality data annotation into a bounty-hunter-style game.
---
1. The Rise of Data Annotation Bounty Hunters
Why Data Quality Matters More Than Compute
Industry consensus in 2024 is clear:
The bottleneck for large AI models is no longer computing power, but high-quality data:
- In fields like programming, law, and healthcare, annotation has evolved into intellectual work requiring reasoning ability, deep expertise, and structural thinking.
- The quality of training data now sets the performance ceiling for AI models.
Example: Surge AI hires top experts — constitutional lawyers with Supreme Court or DOJ experience, medical researchers with peer review capabilities, and rare language specialists.
Hourly rates for such professionals can reach $500–$1,000.
But there’s a critical, unmet demand in AI training:
High-quality software engineering data that captures an engineer’s reasoning process, beyond syntax parsing and code completion.
> This data is scarce, hard to fake, and requires real-world expertise.
---
2. Datacurve’s Distinct Approach: Gamified Annotation
Unlike Surge AI’s traditional outsourcing model, Datacurve built Shipd — a platform that packages engineering tasks as “Quests”:
- Algorithmic problems
- Debugging tasks
- Code comprehension
- Test case creation
- Full repository code reviews
Tasks come with clear prices, and engineers earn cash by completing them.
Quality Control Workflow
Datacurve ensures data integrity through a three-layer validation process:
- Automatic AI validation of submitted work
- Peer code reviews by other engineers, earning separate rewards
- Expert human review for unresolved issues
This “solve → find errors → review” loop maximizes scale without sacrificing quality.
Engineer Earnings Example:
- Rewards range from $80–$100 per task
- Active users have earned $132 in just three days
- High-value tasks reward $250–$350
---
3. Shipd: A Competitive Engineering Arena
Shipd has attracted 16,000+ engineers from companies like:
- Amazon
- AMD
- DeepMind
- OpenAI
- Anthropic
- Vercel
It’s not just about money — Shipd creates a challenge-driven environment resembling an arena, where prestige and achievement are key motivators.
Within two months, Datacurve:
- Surpassed $1M in revenue
- Became a data supplier to Cohere and Anthropic
- Signed the largest contract in its history
---
4. Treating Data Labeling Like a Consumer Product
Core Differentiators
- Platform logic, not manpower scaling — Datacurve manages 10,000+ engineers with a team of fewer than 10.
- Engineers are users, not contractors — Shipd feels like a skill arena, not an outsourcing gig.
- Low marginal costs — validation and scoring are heavily automated.
Co-founder Serena Ge calls it turning data production into a consumer experience, similar to gaming or open-source contributions.
---
5. Funding and Growth
In just one year:
- Seed round: $2.7M
- Series A: $15M
- Total funding: $17.7M
Investor profile:
- Chemistry VC
- Y Combinator
- Backers from DeepMind, OpenAI, Anthropic
Why Investors Are Excited
- Datacurve is filling an expert data gap
- Potential for exponential growth using an internet-product model
- Functions like new AI infrastructure, continuously attracting top-tier professionals
---
6. Industry Implications
The focus of AI development is shifting from compute to ongoing access to high-quality human reasoning.
Datacurve's proposition:
Merge the engineering community with data infrastructure into a new industrial system.
---
7. Related Innovation: AiToEarn
Similarly, platforms like AiToEarn官网 reimagine contribution and monetization:
- Open-source and globally accessible
- AI-powered generation tools
- Integrated cross-platform publishing (Douyin, Bilibili, YouTube, Instagram, Twitter, etc.)
- Analytics and model ranking (AI模型排名)
AiToEarn demonstrates how structured, scalable platforms can empower contributors while keeping operational costs low — much like Datacurve’s engineering arena model.
---
In summary:
Datacurve is not just another annotation company — it’s a gamified, scalable platform producing scarce, high-quality programming data, backed by investor confidence and a growing global engineer community.
---
Would you like me to also create a visual diagram mapping Datacurve’s “solve → find errors → review” process so readers can instantly grasp the closed-loop workflow from this Markdown?
