Joyce K

Content Developer

Jan 2026 - Present • 4 mos

(Working at Correlation One) Content Developer: Machine Learning Data Annotation Responsibilities: • Designing and developing instructional lessons and complete training content packages for data annotation across; text, image, video, audio, and cross-modal modalities. • Producing full session materials including facilitator guides, learner handouts, walkthroughs, practice activities, rubrics, and knowledge checks. • Creating hands-on exercises in Label Studio, including task setup guidance, labeling instructions, and edge case documentation. • Translating complex annotation concepts into clear, learner-friendly lessons with concrete examples accessible to diverse audiences. • Collaborating with program teams in weekly meetings to align on objectives, standards, and delivery timelines; iterate content based on peer and stakeholder feedback. • Supporting content deployment into the company's learning management platform.

Data Annotator

Nov 2025 - Present • 6 mos

(Seasonal Work at iMerit) QA Image Tagging Analyst: - Quality Control & Validation (Auditing): Reviewing annotated images (bounding boxes) and image descriptions generated by AI to detect errors and correcting the errors. - Adherence to Guidelines: Reviewing work against complex, project-specific annotation ontologies and guidelines to ensure compliance. - Edge Case Analysis: Analyzing difficult or unclear images and made judgment calls on how to handle exceptions, ensuring the AI model learns correct, complex scenarios. - Guideline Refinement: Collaborating with Project Managers to suggest improvements to labeling instructions based on common errors observed. - Performance Metrics Tracking: Monitoring and reporting on accuracy metrics and identified trends in errors to improve the overall quality of the dataset. Prompt Evaluation: - The goal of this project was to evaluate the quality of AI-generated images. - First, determining which image in a pair is of lower quality based on prompt adherence and visual integrity. - Then, identifying the specific errors which influenced your decision. - The tool used for annotation was a customized tool (Ango). Image Quality Rater: - Rating AI generated images and categorize the worst and the best to help refine AI generated images. - This role sharpened my eye on details the ordinary eye might miss.

Data Annotator

Oct 2025 - Present • 7 mos

(Working at Innodata) 1. Data Annotation & Quality Assurance- Executing high-precision data labeling and annotation for large-scale datasets, ensuring 99%+ accuracy in alignment with Meta’s rigorous project taxonomies and quality standards. 2. Model Evaluation & RLHF- Performing Reinforcement Learning from Human Feedback (RLHF) by ranking model-generated responses, directly contributing to the fine-tuning of Large Language Models (LLMs) for improved factual accuracy and reasoning. 3. Linguistic & Semantic Analysis- Analyzing complex linguistic patterns and semantic nuances to identify and mitigate model hallucinations, ensuring outputs remained contextually relevant and safe for diverse global audiences. 4. Edge Case Identification- Identifying and documenting edge cases in model performance, providing actionable feedback to engineering teams to improve the system's handling of ambiguous or adversarial prompts. 5. Technical Documentation & Guidelines- Collaborating on the iteration of internal project guidelines, streamlining task workflows and assisting in the calibration of evaluation metrics to increase overall team throughput.