For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
Lily Bonesso

Lily Bonesso

Writer & Creative Producer - Prompt Engineering & Visual Data

United Kingdom flagLondon, United Kingdom
$20.00/hrEntry LevelOther

Key Skills

Software

Other

Top Subject Matter

Fashion / E-commerce – Product Description, Visual Annotation & Image Understanding
Language & Writing – Native English, Editing, Prompt Writing (SFT)
Photography & Video – Visual Analysis, Prompt Writing (SFT), Image/Video Description & Detection

Top Data Types

ImageImage

Top Task Types

Bounding Box
Object Detection
Text Generation
Question Answering
Transcription
Data Collection
Prompt Response Writing SFT

Freelancer Overview

Founder and Creative Director. Brings 15+ years of professional experience across complex professional workflows, research, and quality-focused execution. As an experienced writer, with an English degree from University College London (2014), I can contribute detailed written feedback, instructions and dialogue, as well as attention to detail across all tasks.

Entry LevelEnglishItalianFrench

Labeling Experience

Aether Task: Video Prompt Writing

VideoPrompt Response Writing SFT
I took part in a video-based AI training task focused on prompt-response writing for supervised fine-tuning. The task involved selecting short video clips from instagram and constructing prompt dialogues that described and represented the visual content. Prompts ranged from simple descriptions to more detailed, instruction-style outputs, to mimic the process of asking an AI to create a video. I used clear, precise language to reflect the details in the videos, supported by frame-by-frame notes to ensure accuracy and completeness.

I took part in a video-based AI training task focused on prompt-response writing for supervised fine-tuning. The task involved selecting short video clips from instagram and constructing prompt dialogues that described and represented the visual content. Prompts ranged from simple descriptions to more detailed, instruction-style outputs, to mimic the process of asking an AI to create a video. I used clear, precise language to reflect the details in the videos, supported by frame-by-frame notes to ensure accuracy and completeness.

2025 - 2026

Aether Task – Image Recognition using Bounding Boxes

ImageBounding Box
I had to help train AI to understand the boundaries of individual objects in a busy photograph, for example a beach scene, or a kids bedroom with lots of toys. I had to draw clear bounding boxes around objects, or redraw inaccurate boxes that had been produced by the AI. In some tasks I also had to write feedback or explain what should be improved. In all tasks I took careful instructions and followed the task criterea.

I had to help train AI to understand the boundaries of individual objects in a busy photograph, for example a beach scene, or a kids bedroom with lots of toys. I had to draw clear bounding boxes around objects, or redraw inaccurate boxes that had been produced by the AI. In some tasks I also had to write feedback or explain what should be improved. In all tasks I took careful instructions and followed the task criterea.

2025 - 2026

Aether Task: Colour Extraction & Image Description

ImagePrompt Response Writing SFT
The task involved uploading and analysing photographs and artworks, with an emphasis on high quality images across a variety of subjects, in order to create a large dataset. Using a colour-dropper tool I had to select the key colours which made up the image. The tool would then identify the key colour values of each colour selected, producing a unique code for that colour. These codes were then incorporated into detailed, structured text descriptions I wrote about the visual content. When describing the image, I had to pay careful attention to nuance in tone, lighting, and artistic variation, tailoring the text to carefully indentify its most defining details.

The task involved uploading and analysing photographs and artworks, with an emphasis on high quality images across a variety of subjects, in order to create a large dataset. Using a colour-dropper tool I had to select the key colours which made up the image. The tool would then identify the key colour values of each colour selected, producing a unique code for that colour. These codes were then incorporated into detailed, structured text descriptions I wrote about the visual content. When describing the image, I had to pay careful attention to nuance in tone, lighting, and artistic variation, tailoring the text to carefully indentify its most defining details.

2025 - 2025

Aether Task: Showing an Scene from Multiple Angles

ImageData Collection
I took part in a data collection project aimed at creating large dataset for AI scene expansion. The task involved sourcing image sets imported from Instagram image carousels to capture the same building, street, or urban environment from multiple angles. This required identifying visually consistent yet varied images that would expand AI's understanding of one scene. I focused on curated architecture and design accounts, as these show a large variety of interesting three-dimensional spaces and forms within buildings and urban landscapes.

I took part in a data collection project aimed at creating large dataset for AI scene expansion. The task involved sourcing image sets imported from Instagram image carousels to capture the same building, street, or urban environment from multiple angles. This required identifying visually consistent yet varied images that would expand AI's understanding of one scene. I focused on curated architecture and design accounts, as these show a large variety of interesting three-dimensional spaces and forms within buildings and urban landscapes.

2025 - 2025

Xylophone Task: Speaking in Native English

AudioQuestion Answering
In order to train AI to understand and replicate native English speaking language, including local dialect in Britain and the USA, we were tasked with a series of zoom calls, which could be recorded in order to create large datasets. For each zoom call we had to select a topic and speak exclusively about this topic, using clear language, not speaking over each other and contributing a range of opinions to add breadth to the subject matter.

In order to train AI to understand and replicate native English speaking language, including local dialect in Britain and the USA, we were tasked with a series of zoom calls, which could be recorded in order to create large datasets. For each zoom call we had to select a topic and speak exclusively about this topic, using clear language, not speaking over each other and contributing a range of opinions to add breadth to the subject matter.

2025 - 2025

Education

U

University College London

Bachelor of Arts, English Literature

Bachelor of Arts
2011 - 2014
K

Kingston University

Foundation Diploma in Art and Design, Art Foundation

Foundation Diploma in Art and Design
2010 - 2011

Work History

H

Hyfae Shop & Magazine

Founder and Creative Director

London
2024 - Present
F

Freelance

Producer, Director and Writer

London
2023 - Present