Prompt Engineer - Health AI Classification (95%+ Accuracy)

Remote | Hourly Contract | Potential full-time

We're launching supplement analysis platform for 25,000+ of products but accuracy of data isn't good enough. Infrastructure is done. Prompts need fixing.

What you'll do:

• Optimize 3 critical LLM classification tasks (data: filtering, categorization, extraction)
• Achieve >95% accuracy with cost-effective models (e.g. Gemini, Grok)
• Build regression test coverage in external testing enviroment
• Analyze failure patterns and improve prompts systematically
• Document methodology for team

Requirements:

• Proven track record achieving >95% accuracy on complex classification
• Experience with high-stakes AI (healthcare/finance/legal)
• Systematic testing approach (not trial-and-error)
• Multi-model expertise (GPT-4, Claude, Gemini, etc.)

Nice to have:

• Healthcare/supplement domain knowledge
• MySQL or similar familiarity
• Structured output optimization (JSON mode, function calling)
• Fine-tuning experience

We provide:

• ⁠Complete infrastructure (external testing enviroment, admin panels)
• Multi-LLM access (Claude, GPT, Gemini, Grok, OpenRouter)
• Data analyst for verification support
• Clear success metrics (>95% accuracy)
• Competitive hourly rate
   

To Apply:
•⁠  Write us message by clicking "Apply for the job"
•⁠  Fill out questioner here: https://tally.so/r/mBO904