Every project here solved a business problem and has had measurable impact on operations.
A company was using Temi and a human transcript reviewer to transcribe and clean hours of interviews per day. The process was slow, expensive, and error-prone
I took an employee's transcription app and added AI-powered cleaning. I optimized the code for efficiency and security, then deployed it as an internal web app.
80%+ cost savings; 90%+ reduction in processing time; errors reduced significantly
No easy way to empirically compare LLM configurations for a specific task
Experiment runner with side-by-side output comparison and AI-generated analysis
Teams use the correct model for each use case, reducing errors and optimizing costs
Style rules weren't being followed and content wasn't inserted into templates correctly
Structured, multi-phase pipeline with audit phases and human-in-the-loop oversight
80%+ reduction in content drafting and templating time