Senior Full-Stack AI/ML Engineer (Data Labeling & AI Training)
Document
Fine-tuningDesigned training-data pipelines that transform raw content assets into annotation-ready structured inputs for consistent supervision. Implemented labeling evaluation loops and dataset validation gates to ensure quality, schema consistency, and distribution stability prior to model consumption. Operationalized dataset-driven benchmarking and monitoring to quantify the impact of improved labeled data and trigger retraining when drift is detected. • Built RAG ingestion workflows that extract metadata and layout-aware features to support accurate labeling and high-fidelity supervision • Prepared LLM fine-tuning datasets using LoRA/QLoRA and aligned labeled outputs to safety constraints, brand voice, and compliance requirements • Implemented human/SME evaluation UI to review model outputs and generate high-quality labeled feedback for iteration • Deployed drift monitoring and auditability instrumentation using EvidentlyAI and Evidently-driven tests to detect distribution shifts early.