Gaser Elmasry

Key Skills

Software

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Image

Text

Top Task Types

Computer Programming/Coding

Evaluation/Rating

Prompt + Response Writing (SFT)

RLHF

Text Generation

Freelancer Overview

As a Generative AI Specialist, I focus on enhancing and evaluating premier Large Language Models, including Claude 3.7, Gemini, and Grok. My expertise lies in LLM evaluation, Reinforcement Learning from Human Feedback (RLHF), and advanced prompt engineering. I specialize in meticulously refining model responses, correcting complex generated code, and developing ideal solutions across a variety of programming languages like Python, C++, and Java to ensure the highest standards of accuracy and performance. My approach involves engineering sophisticated adversarial prompts and creating comprehensive "golden" test suites to rigorously test model limitations and identify critical failure points. This work directly contributes to tangible improvements in model robustness and reliability. I have practical experience fine-tuning LLMs on complex, real-world open-source projects like DVC and Moto. My consistent high-quality contributions led to my recognition as a top-tier contributor and an invitation to a leadership role as a "Coding Network Lead."

Labeling Experience

Generative AI Specialist: LLM Evaluation and Code Refinement

Scale AIComputer Code ProgrammingRLHFFine Tuning

Enhanced and evaluated premier Large Language Models, including Claude 3.7, Gemini, and Grok. My primary tasks involved refining model-generated text, correcting complex code, and developing ideal solutions across multiple languages (C, C++, Java, Python). I engineered sophisticated and adversarial prompts to test model limitations and improve robustness. Executed over 150 tasks across 18+ distinct projects, consistently adhering to strict quality measures to ensure the highest standards of accuracy and performance.

Enhanced and evaluated premier Large Language Models, including Claude 3.7, Gemini, and Grok. My primary tasks involved refining model-generated text, correcting complex code, and developing ideal solutions across multiple languages (C, C++, Java, Python). I engineered sophisticated and adversarial prompts to test model limitations and improve robustness. Executed over 150 tasks across 18+ distinct projects, consistently adhering to strict quality measures to ensure the highest standards of accuracy and performance.

2024

Education

Work History