London, United Kingdom
+44 7493 513 169
graemeweeden@croftlandsconsulting.com
Graeme Weeden — Senior Finance Leader & AI Transformation Specialist
Graeme Weeden is a Chartered Accountant with over 20 years of experience operating at board and executive level across private equity-backed, acquisitive and complex multi-entity businesses.
His career began at KPMG, where he qualified as a Chartered Accountant before moving into investment banking holding roles as Head of Operations, Head of Market Risk, CFO and Head of Equities Transformation
Since 2010, Graeme has operated primarily as a finance consultant and interim and fractional CFO to SMEs and growth businesses, with recent senior engagements at Dext Software (global PE-backed SaaS), Informa plc (FTSE 100) and Raccoon Media Group (PE-backed international events).
Graeme is also a graduate of the MIT Professional Education Applied AI & Data Science Program (2026 cohort). The programme spanned eight courses: foundations in Python and statistics; data analysis and visualisation; machine learning; practical data science; deep learning and neural networks; computer vision; natural language processing and generative AI; and recommendation systems and agentic AI.
The MIT Programme — The Eight Courses
The MIT Applied AI & Data Science Program is an eight-course, hands-on curriculum taught by MIT faculty including Prof. Caroline Uhler (unsupervised learning and network analysis), Prof. Munther Dahleh (ensemble methods and time series) and Prof. Devavrat Shah (recommendation systems). The eight courses:
The core toolkit for working in Python with Pandas and NumPy, and covering descriptive and inferential statistics, probability distributions, and hypothesis testing.
Exploratory analysis and techniques for making sense of large, high-dimensional datasets, including dimensionality reduction with PCA and t-SNE, and unsupervised clustering.
Supervised learning for prediction, covering both regression and classification, and how to evaluate a model properly with cross-validation rather than trusting a single result.
The methods used most in real analysis, including decision trees, ensemble models, and time series forecasting.
How neural networks learn and where they outperform traditional models, building and training networks on real data.
Using convolutional neural networks to interpret images, covering image classification and transfer learning.
Working with text data, from language fundamentals through to large language models, prompt engineering, and retrieval-augmented generation for grounding AI answers in a company’s own documents.
Building systems that personalise and recommend, and designing autonomous AI agents that carry out multi-step tasks with limited supervision.