Vikash kumar

I’m a remote AI data contributor with hands-on experience in structured image description, bounding box and polygon annotation, visual content labeling, and AI output evaluation. I’m Hindi–English bilingual with professional English fluency, and I consistently deliver annotation-ready work at volume with no supervision required.

In my contract work (2024–Present), I produced precise, detail-rich written descriptions of photographs, diagrams, and illustrations using bounding box, polygon, and keypoint annotation techniques. By applying systematic observation frameworks and strict quality checklists, I achieved a 98%+ QA pass rate across 500+ evaluated and labeled assets.

I also improve AI training data quality by identifying and correcting visual and numerical discrepancies in complex informational datasets. Using inter-annotator quality scoring and RLHF-aligned review standards, I reduced error rates by 30%+ across each review cycle through pattern recognition and logical cross-referencing.

To support quantitative research workflows, I deliver semantic segmentation and LLM evaluation outputs with zero ambiguity by structuring tasks around explicit parameters and client-defined quality benchmarks. I maintain strong rubric-based communication, adhere to style guides, and complete digital asset creation tasks using Canva and text customization platforms while performing visual QA and content moderation checks.