Soha Ashraf

Key Skills

Software

Clickworker

Labelbox

Remotasks

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Image

Text

Top Task Types

Bounding Box

Classification

Computer Programming Coding

RLHF

Text Summarization

Freelancer Overview

As an AI Engineer and RLHF Specialist at Scale AI, I played a pivotal role in training and refining AI models through meticulous data labeling, prompt engineering, and iterative model evaluation. My work involved designing and optimizing prompts to enhance model responses, rigorously assessing output quality, and implementing feedback loops to improve performance. I collaborated cross-functionally to ensure alignment with project goals, leveraging my expertise in reinforcement learning and human feedback (RLHF) to fine-tune models for accuracy and relevance.

Labeling Experience

Data Labeling for AI Training Text, and Images

Scale AIImageClassificationQuestion Answering

Rated AI model performance on extracting accurate information from images and answering related prompts (e.g., "What's in this image?" or "Describe this scene"). Evaluated visual grounding (how well text responses matched image content). Flagged hallucinations or incorrect details in model outputs.

Rated AI model performance on extracting accurate information from images and answering related prompts (e.g., "What's in this image?" or "Describe this scene"). Evaluated visual grounding (how well text responses matched image content). Flagged hallucinations or incorrect details in model outputs.

2024

AI Model Training & RLHF Data Labeling for Programming Tasks

Scale AIComputer Code ProgrammingClassificationRLHF

As a Data Labeler and AI Trainer for Scale AI, I specialized in refining AI models for programming-related tasks. My work included: Labeling and annotating code snippets, prompts, and model responses to improve accuracy and relevance. Evaluating and rating AI-generated code outputs based on correctness, efficiency, and adherence to instructions. Designing prompts and fine-tuning responses for supervised fine-tuning (SFT) to enhance model performance. Implementing RLHF techniques to iteratively improve model behavior through human feedback loops. Collaborating with cross-functional teams to align labeling guidelines with project goals, ensuring high-quality training data. This project required deep familiarity with programming languages ) and AI/ML principles, enabling me to deliver precise annotations and actionable feedback for model optimization.

As a Data Labeler and AI Trainer for Scale AI, I specialized in refining AI models for programming-related tasks. My work included: Labeling and annotating code snippets, prompts, and model responses to improve accuracy and relevance. Evaluating and rating AI-generated code outputs based on correctness, efficiency, and adherence to instructions. Designing prompts and fine-tuning responses for supervised fine-tuning (SFT) to enhance model performance. Implementing RLHF techniques to iteratively improve model behavior through human feedback loops. Collaborating with cross-functional teams to align labeling guidelines with project goals, ensuring high-quality training data. This project required deep familiarity with programming languages ) and AI/ML principles, enabling me to deliver precise annotations and actionable feedback for model optimization.

2024

Education

Work History