Fatima B.

Key Skills

Software

Don't disclose

Other

Top Subject Matter

Speech recognition data preparation (Urdu/Roman Urdu)

Computer vision training data (action-based videos)/AI TRAINING

RHLF,MTPE,DATA LABELLING,BETA TESTING

Top Data Types

Audio

Video

Text

Image

Top Task Types

Audio Recording

Action Recognition

RLHF

Entity (NER) Classification

Freelancer Overview

Linguistic Recordings for Acoustic Modeling (Urdu Voice Data Collection). Core strengths include Don't disclose and Other. Education includes Bachelor of Science, University of Faisalabad (TUF) (2028). AI-training focus includes data types such as Audio, Video, and Text and labeling workflows including Audio Recording, Action Recognition, and Translation.

Labeling Experience

Data Labeling & Semantic Annotation for NLP Logic

Don't discloseTextEntity (NER) Classification

Performed semantic annotation by categorizing complex datasets for machine learning model training. Supported NLP logic-oriented dataset structuring through careful label assignment. Focused on producing consistent, high-quality annotations to improve downstream NLP performance. • Semantic annotation and categorization • Dataset labeling for NLP model training • Consistency-focused annotation • Support for logic-based NLP systems

2026 - Present

Internet Judging / Search Evaluation (Relevance & Map Accuracy)

Don't discloseText

Evaluated search query relevance and map accuracy as part of internet judging and search evaluation style tasks. Produced quality UX feedback and search evaluation signals to help improve AI and retrieval performance. Applied criteria-based assessment to support dataset labeling for ranking or relevance. • Search query relevance evaluation • Map accuracy judging • UX feedback generation for improving products • Criteria-based evaluation for dataset usefulness

2026 - Present

LLM Urdu Tutoring + RLHF/Prompt Tuning Support

Don't discloseTextRLHF

Contributed human feedback workflows intended for RLHF and prompt tuning to align AI responses with user intent and safety guidelines. Supported training for Urdu language performance in LLMs by providing tutoring and guidance during AI alignment. Focused on improving model helpfulness and linguistic suitability for Urdu and Roman Urdu contexts. • RLHF-style human feedback for response alignment • Prompt tuning to meet intent/safety constraints • Urdu AI tutoring for LLM behavior • Feedback-driven improvement for model quality

2026 - Present

MTPE and Localization / Roman Urdu Specialist (Text Work)

OtherText

Performed Roman Urdu specialization work to adapt and localize content for the Pakistani digital market. Completed MTPE-style refinement of machine-translated Urdu text for technical precision and natural flow. Supported localization efforts by ensuring cultural and technical accuracy for Urdu audiences. • Roman Urdu content adaptation • MTPE (machine translation post-editing) for Urdu • Localization for cultural/technical correctness • Terminology and style consistency for UX-focused outputs

2026 - Present

Video Annotation/Data Collection for Action-Based Vision Models

Don't discloseVideoAction Recognition

Collected and prepared action-based video data intended for computer vision training and annotation pipelines. Captured video content such as hand gestures to support vision models. Applied consistent filming and dataset preparation practices to improve training usefulness. • Recording action-based videos for AI vision training • Capturing hand-gesture and gesture-based actions • Dataset preparation for computer vision learning • Quality-focused recording environment

2026 - Present

Show more

Education

Work History