UBER Case Study
How DataAlchemy.AI Helped Uber
Uber aimed to fine-tune its LLMs for Finance, Energy, and Manufacturing but struggled with thousands of French PDFs requiring precise query generation and relevance ratings. DataAlchemy.AI stepped in to deliver bilingual, accurate, and consistent annotations at scale.
150K+ Precise Annotations
22% Accuracy Lift
95% Client
CHALLENGES
Uber’s team came to us with several key hurdles:
Here’s how we tackled it together:
Experts
Handpicked native French bilingual experts with deep domain knowledge, so each annotation was precise and context-aware.
Guidelines
Created detailed, standardized annotation guidelines to remove ambiguity and keep everyone on the same page.
Reviews
Built a multi-layer review process—annotator → reviewer → auditor—to catch errors early and ensure quality.
Formats
Formatted deliverables in LLM-ready JSON/CSV and worked closely with Uber SMEs to make sure every annotation aligned with the intended domain context.
CONCLUSION
By blending domain expertise, careful processes, and close collaboration, we delivered over 150,000 high-quality annotations. This not only improved Uber’s LLM retrieval accuracy by 22% but also earned a client acceptance rate above 95%. More than numbers, this project was about turning a massive, complex challenge into a smooth, confident solution, letting Uber focus on deploying its LLMs effectively across industries.