AI Training Specialist
I worked on large-scale AI training and dataset creation projects focused on improving the accuracy, reasoning, and real-world performance of advanced AI systems. The scope of work included creating, reviewing, and refining high-quality training data for software engineering, terminal workflows, and web application scenarios. My responsibilities involved benchmarking AI outputs, labeling correct and incorrect responses, writing step-by-step technical solutions, debugging guides, and evaluating multi-turn agent behavior. I contributed to Terminal Bench, SWE Bench, and RLHF workflows used by enterprise-level clients, ensuring that the data closely matched real engineering use cases. The projects were high-volume in nature, with hundreds of tasks completed across multiple datasets. Quality was maintained through strict adherence to client guidelines, detailed justification for labels, preference ranking, peer review, and iterative feedback loops. Accuracy, consistency, and clarity were