Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I’m Vhm Ed Sherif, an AI Trainer and mobile developer with over three years of hands-on experience improving large language model performance through LLM alignment, Reinforcement Learning from Human Feedback (RLHF), and Red Teaming. I also work on multilingual (Arabic and English) evaluation and prompt-and-response training data, with a strong focus on safety, quality, and culturally accurate outputs.…I’m Vhm Ed Sherif, an AI Trainer and mobile developer with over three years of hands-on experience improving large language model performance through LLM alignment, Reinforcement Learning from Human Feedback (RLHF), and Red Teaming. I also work on multilingual (Arabic and English) evaluation and prompt-and-response training data, with a strong focus on safety, quality, and culturally accurate outputs.

Vhm Ed Sherif

Data Scientist, AI Trainer, AI Evaluator, +1





I’m Vhm Ed Sherif, an AI Trainer and mobile developer with over three years of hands-on experience improving large language model performance through LLM alignment, Reinforcement Learning from Human Feedback (RLHF), and Red Teaming. I also work on multilingual (Arabic and English) evaluation and prompt-and-response training data, with a strong focus on safety, quality, and culturally accurate outputs.…I’m Vhm Ed Sherif, an AI Trainer and mobile developer with over three years of hands-on experience improving large language model performance through LLM alignment, Reinforcement Learning from Human Feedback (RLHF), and Red Teaming. I also work on multilingual (Arabic and English) evaluation and prompt-and-response training data, with a strong focus on safety, quality, and culturally accurate outputs.

Available to hire

I’m Vhm Ed Sherif, an AI Trainer and mobile developer with over three years of hands-on experience improving large language model performance through LLM alignment, Reinforcement Learning from Human Feedback (RLHF), and Red Teaming. I also work on multilingual (Arabic and English) evaluation and prompt-and-response training data, with a strong focus on safety, quality, and culturally accurate outputs.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Language

Arabic

Fluent

English

Advanced

Work Experience

Freelance AI Trainer & Linguistic Evaluator at Invisible Technologies | SuperAnnotate | Productive Playhouse | Outlier

January 1, 2021 - Present

Conduct adversarial testing (red teaming) by authoring complex, malicious prompts to evaluate LLM safety guardrails, discover vulnerabilities, and test model boundaries. Design multimodal evaluation tasks (OTS) pairing intricate image data with detailed text prompts to benchmark and fine-tune vision-language model responses. Perform supervised fine-tuning (SFT) by developing high-quality prompt-and-response training datasets to help models exceed baseline reasoning capabilities. Perform audio quality assurance by evaluating audio training data and synthesized speech quality, focusing on background noise isolation and acoustic calibration.