Preyash Kharva

I am a detail-oriented AI tutor and data specialist with hands-on experience training and evaluating large language models such as Gemini and GPT. I focus on annotation, data labeling, and quality assurance across text, image, and multimodal datasets to improve model comprehension and safety.

At NamanSoft I progressed from Data Analyst to Generative AI Engineer, collaborating with product and engineering teams to produce high-quality datasets for fine-tuning and to evaluate model accuracy and conversational naturalness. I act as a bridge between engineers and annotators to ensure labeling precision and consistency.

In roles with Outlier AI, Handshake AI (MOVE Fellowship), and Abaka AI I performed instruction-following evaluations, developed rubrics for safe conversational boundaries, and audited complex linguistic structures to enhance alignment, factuality, and bias mitigation. I provide structured scoring and linguistic insight to refine model outputs.

I also build applied projects and automation solutions — for example, an AI voice-and-tool assistant integrating GPT, ElevenLabs, and n8n, and automation tooling using Python and SQL for data reporting. I hold multiple generative AI and cloud certifications and seek hybrid roles where I can advance model safety, data quality, and conversational performance.