Zara J.
Mid-Level AI Data Annotation Specialist
Open to offersNew to PlatformAbout
Zara J. is an accomplished AI Data Annotator & RLHF Specialist with extensive experience in training Large Language Models (LLMs) and computer vision systems. Demonstrating expertise in Reinforcement Learning from Human Feedback (RLHF) and complex data annotation, Zara excels in bridging the gap between raw data and machine intelligence. Her proficiency includes evaluating multi-turn conversational AI models to boost accuracy and adherence to constraints, as well as auditing AI outputs to eliminate machine hallucinations. With a proven track record in prompt engineering, she skillfully develops adversarial prompts to test AI safety guardrails. Zara's hands-on experience in semantic segmentation and multi-object tracking has seen her label over 50,000 high-resolution images, maintaining a 99.4% accuracy score, often by adapting to evolving project requirements. Trusted by startups for her fast-checking abilities, she ensures that datasets meet stringent quality metrics.
Experience
AI Training & RLHF Specialist
Contract · 2025 — PresentEvaluated and ranked thousands of model-generated responses for multi-turn conversational AI projects, enhancing model helpfulness, accuracy, and compliance with constraints. Conducted audits of high-complexity AI outputs, including technical reasoning and factual summaries, against authoritative sources to eliminate machine hallucinations. Developed diverse adversarial prompts intended to stress-test safety guardrails and identify structural vulnerabilities in models.Data Annotation Specialist
Outsourced Contract · 2024 — 2024Labeled over 50,000 high-resolution images and video frames for computer vision models, focusing on semantic segmentation and multi-object tracking. Achieved a consistent consensus/accuracy score of 99.4%, often promoted to Reviewer to audit the work of junior annotators. Adapted quickly to evolving labeling taxonomies across various simultaneous projects, including those for autonomous vehicles and e-commerce categorization.