AI Data Engineer / LLM Trainer (Contract) — Andela (Remote)
Generated high-quality multi-turn conversational datasets for LLM fine-tuning by simulating realistic user–AI interactions. Designed dialogues that incorporate JSON-formatted function-calling tool use to support multi-step reasoning scenarios. Performed iterative review and quality validation of 100+ dialogue examples weekly to meet internal scoring rubrics.• Multi-turn conversation dataset creation for assistant training across productivity and domain-specific tools• User/assistant role emulation with logical function-calling represented as JSON API calls• Agent-completion dataset creation including feasible task resolution and graceful infeasibility handling• Automated Python-based JSON data validation and quality checks prior to submission