AI Challenge Competitor, Zindi African Trust & Safety LLM Challenge
TextRed Teaming
Participated in an adversarial prompt engineering challenge targeting multilingual African LLMs. The main goal was to identify trust-and-safety failure modes in large language models using diverse linguistic inputs. Developed, evaluated, and submitted structured prompts designed to test model robustness. • Crafted adversarial prompts across multiple languages including English, Swahili, and Hausa.
• Evaluated LLM trust and safety issues using specific red teaming methodologies.
• Collaborated in a competitive AI training environment focused on real-world risk assessment.
• Submitted a final set of prompts for formal evaluation and feedback.
Participated in an adversarial prompt engineering challenge targeting multilingual African LLMs. The main goal was to identify trust-and-safety failure modes in large language models using diverse linguistic inputs. Developed, evaluated, and submitted structured prompts designed to test model robustness. • Crafted adversarial prompts across multiple languages including English, Swahili, and Hausa. • Evaluated LLM trust and safety issues using specific red teaming methodologies. • Collaborated in a competitive AI training environment focused on real-world risk assessment. • Submitted a final set of prompts for formal evaluation and feedback.
2024 - 2024