Data Scientist - Image Conversion Probability

Freelance Lead Data Scientist
Start: Within 1–2 weeks)
Contract: 1 month (strong potential to extend – Inside preferred for extensions)
Day Rate: £600/day (Inside or Outside IR35 considered)
Location: Remote (UK-based preferred; preferred onsite for initial few days
We’re supporting a major client on a short, high-impact project that blends computer vision, NLP, and marketing analytics. The goal is to refine and optimise an existing model that predicts the conversion rate of marketing images based on their visual and contextual attributes.
The work:

• Enhance or rebuild an existing multimodal image-to-text model built using CLIP and Phi (OpenAI).
• Apply SHAP (or similar explainability methods) to attribute what’s driving conversion — e.g., the presence of people, objects, or scene context.
• Extend an interactive Python Dash web app that visualises image predictions and their drivers.
• Work closely with the analytics and marketing teams to make the tool usable, explainable, and business-ready.
• Handle both model optimisation and dash app refinement, focusing on interpretability and usability.

Tech environment:
Python | CLIP | Phi | SHAP | Computer Vision | NLP | Dash | Pandas | NumPy
Ideal profile:

• Proven experience in data science / ML, with practical exposure to computer vision and explainable AI.
• Strong coding in Python, with experience building models end-to-end and deploying lightweight web apps (Dash/Streamlit).
• Understanding of model explainability (SHAP, LIME, feature attribution).
• Hands-on with multimodal data (images + text) preferred.
• Comfortable in a fast-moving, creative environment; self-starter with a growth mindset.

Contract length: 4–5 weeks initially, strong chance of extension into wider data/AI projects