Ansar Khangeldin

Key Skills

Software

Google Cloud Vertex AI

Toloka

Don't disclose

Top Subject Matter

No subject matter listed

Top Data Types

3D Sensor

Text

Video

Top Task Types

Action Recognition

Computer Programming Coding

Data Collection

Entity Ner Classification

Object Detection

Freelancer Overview

I have hands-on experience in designing high-quality training data for large language models, with a focus on adversarial and reasoning-challenging examples. I crafted over 600 complex questions that targeted reasoning gaps and factual inconsistencies in LLM (NDA), which were later validated and used to improve model performance. I also validated over 300 question-answer pairs to ensure clarity, correctness, and alignment with model capabilities. My process emphasized edge cases, step-by-step reasoning, and evaluation beyond surface-level accuracy — skills critical for robust model improvement. In addition, my background in computer vision research gives me a solid understanding of working with image and video data. While I haven’t done manual labeling, I have practical experience in dataset preparation, pre-processing, and evaluation pipelines — particularly in tasks like object and motion detection. This blend of LLM training data expertise and CV data handling enables me to bridge NLP and vision data quality needs effectively.

Labeling Experience

Education

Work History