Freelance Front-End Developer & AI Data Consultant (AI Model Training & SFT / LLM Evaluation)
AI training tasks centered on creating prompt-response validation datasets to benchmark and improve model comprehension for web/framework-related behaviors. The work included evaluating model outputs using multi-axis criteria such as truthfulness, instruction adherence, formatting constraints, reasoning logic, and safety alignment. Results were used as training fixes and to guide supervised fine-tuning style improvements for LLM performance. • Engineered complex prompt-response validation datasets for model benchmarking • Audited machine-generated React/JavaScript outputs and produced corrected targets as training fixes • Performed multi-axis ranking and QA over thousands of conversational AI examples • Wrote thousands of analytical justification/rationale descriptions for acceptance and rejection decisions