Muhammad Umair

Key Skills

Software

AWS SageMaker

Appen

CrowdSource

Mindrift

OneForma

Top Subject Matter

No subject matter listed

Top Data Types

Document

Image

Text

Top Task Types

Data Collection

Prompt Response Writing SFT

Question Answering

Text Generation

Translation Localization

Freelancer Overview

As a multilingual linguist and AI training specialist, I bring over five years of experience in translation, proofreading, post-editing of machine translation (MTPE), and text data labeling. I have worked on diverse NLP and LLM evaluation projects involving English, Spanish, Arabic, Hindi, and Urdu, contributing to the improvement of AI systems' language understanding, accuracy, and user intent alignment. I specialize in labeling and annotating text data, evaluating AI-generated content for factual consistency, and crafting/refining prompts to improve model performance. My background includes successful work with platforms like Translate.com, Appen, and various localization companies, where I provided high-quality linguistic QA and content moderation services. With a strong command of CAT tools such as SDL Trados, MemoQ, and MateCat, I ensure quality, consistency, and efficiency in every task. My keen eye for detail, cross-linguistic knowledge, and commitment to excellence make me a reliable contributor to AI data training and language projects.

Labeling Experience

Medical Text Annotation Project – NER & Text Classification

AppenTextEntity Ner ClassificationTranslation Localization

Worked on a large-scale text annotation project focused on Named Entity Recognition (NER) and text classification for training multilingual AI models. The task involved labeling entities such as person names, organizations, locations, and medical terms in English, Urdu, and Arabic datasets. The project required consistent adherence to detailed annotation guidelines and weekly quality checks. I processed over 10,000+ sentences using Labelbox, maintaining over 97% accuracy, verified through peer and automated QA tools. Regular feedback sessions were conducted with project managers to ensure consistency and performance optimization.

Worked on a large-scale text annotation project focused on Named Entity Recognition (NER) and text classification for training multilingual AI models. The task involved labeling entities such as person names, organizations, locations, and medical terms in English, Urdu, and Arabic datasets. The project required consistent adherence to detailed annotation guidelines and weekly quality checks. I processed over 10,000+ sentences using Labelbox, maintaining over 97% accuracy, verified through peer and automated QA tools. Regular feedback sessions were conducted with project managers to ensure consistency and performance optimization.

2024 - 2024

Education

Work History