LLM Engineer – AI Training/Data Labeling and Evaluation
Text
RLHFAs an LLM Engineer at DataCurve, I contributed to AI training by enhancing proprietary and open large language models. My tasks included orchestrating multiple LLMs, building retrieval-augmented generation pipelines, and applying RLHF and RLMF to generate high-quality datasets for training workflows. I also designed real-time problem generation systems to challenge and improve reasoning models under edge cases. • Generated FFT datasets for NVIDIA model training leveraging reinforcement learning from human feedback. • Applied model red-teaming to test robustness and reasoning of advanced LLMs (e.g., Claude). • Developed RLHF pipelines to improve the contextual understanding of LLMs. • Orchestrated and evaluated the performance of multiple LLMs, ensuring continuous improvement.