Hire this AI Trainer
Sign in or create an account to invite AI Trainers to your job.
ClickworkerData Annotation TechOneFormaRemotasksScale AIV7 LabsLabelboxMindriftToloka
CVATSurge AI
Audio
Computer Code Programming
Document
Image
Text
VideoI am an experienced AI training data specialist with a background in data annotation, linguistic evaluation, and prompt engineering for large language models. My work includes collecting and labeling diverse image and video datasets, evaluating AI assistant performance, and refining prompts to enhance NLP applications. I have hands-on experience with tools for chatbot development, model fine-tuning, and A/B testing, as well as expertise in Python, JavaScript, and SQL. My academic foundation in AI and software development, combined with practical projects in e-commerce, computer vision, and cybersecurity, enables me to deliver high-quality, accurate training data and insightful feedback to optimize machine learning models. I thrive in multicultural teams and am committed to clear communication and continuous improvement in AI systems.
This project involved several media types within different task categories. From labelling with Scale AI tool, to response rating, to prompt evaluation, data collection, mapping, and transcription.
Project Aether is a high-volume AI training project on the Outlier.ai platform, operated by Scale AI. It focuses on human-in-the-loop evaluation and reasoning tasks, such as comparing image accuracy, finding visual discrepancies, and generating short dialogues. It is designed to train large language models (LLMs) and is known for being relatively straightforward and accessible to contributors. Tasks include evaluating visual accuracy, spotting differences in visuals, summarising images/video, and writing natural conversation scripts. It is a high-volume, fast-paced project where workers can sometimes work up to 10 hours a day. Aether emphasises detail-oriented tasks to help AI distinguish between "almost right" and "exactly right". The project involves strict quality requirements, with some reports indicating that in-platform reviews or "purges" are used to maintain high standards or change the focus of the work
Researching and training large language models (LLMs) by creating a scenario-specific prompt, recording it with the intended tone, and rating the model's response based on predefined criteria and dimensions.
The project was aimed to rank model responses using critical and non-critical dimensions and utilising a rejection cheatsheet - to help determine coherent and incoherent responses. The project did not have a criteria group.
The project desirable was to ensure workers delivered high quality diagrams from any STEM background and correctly labelling, correcting and fixing bounding boxes to wrap text without overlapping. The software first detects the text and bounds automatically, but there are inconsistencies a times, hence why workers are required to be very detail oriented enough to find errors and fix them.
MSc, Computer Science
Master of Science, Computer Science
A.I Trainer and Evaluation specialist
Linguistic and LLM Evaluator and Annotator
