Leon Hamnett's Machine Learning and Data Science Portfolio

Leon Hamnett – Machine Learning Engineer, Researcher, and Consultant.

I help climate and sustainability-focused companies turn data into impact through powerful, production-ready machine learning systems.

My past work includes:

• 🌲 Building satellite-based deforestation detection models for Tanzanian forests and mangroves
• ☀️ Developing solar energy forecasting tools for utility-scale solar farms
• 🌊 Predicting hydropower energy generation via rainfall and resrvoir fill rates
• 🔬 Leading medical AI research for cancer segmentation and classification

I offer flexible consulting (hourly or project-based), with services including:

• Custom model development and optimisation
• MVP builds to quickly validate your ideas
• Data pipeline architecture for efficient, scalable workflows
• MLOps integration to automate and deploy your models
• Exploratory data analysis and feature engineering to uncover actionable insights

🔍 Curious about my track record?
Check out my recommendations or scroll down to explore published machine learning research, portfolio projects, and code samples.

LinkedIn: Github:

📜 Machine Learning Research Publications:

Completed research papers:

• Enhancing Transformer-Based Segmentation for Breast Cancer Diagnosis using Auto-Augmentation and Search Optimisation Techniques (Hamnett et al, 2023)
  Accepted for ML4Health conference 2023 (previously part of NeurIPS conference)
  Paper Link
  Code Link

  Abstract Snippet:
  This paper introduces a methodology that combines automated image augmentation selection (RandAugment) with search optimisation strategies (Tree-based Parzen Estimator) to identify optimal values for the number of image augmentations and the magnitude of their associated augmentation parameters, leading to enhanced segmentation performance. We empirically validate our approach on breast cancer histology slides, focusing on the segmentation of cancer cells. A comparative analysis of state-of-the-art transformer-based segmentation models is conducted, including SegFormer, PoolFormer, and MaskFormer models, to establish a comprehensive baseline, before applying the augmentation methodology. Our results show that the proposed methodology leads to segmentation models that are more resilient to variations in histology slides whilst maintaining high levels of segmentation performance, and show improved segmentation of the tumour class when compared to previous research.

Research papers in progress:

• Improved stratified k-fold cross validation methodology for medical domain
  
• Detecting and mitigating racial bias in breast cancer image classification models
  Investigating the presence of racial bias in breast cancer detection models and methods which may be used to reduce the racial bias present in a trained image classifier.

📚 Deep Learning Projects:

Deep Learning Projects Repo

This repository contains a collection of my work building neural networks in the areas of computer vision and NLP. Here's a glimpse of what you'll find:

Computer Vision Projects:

• Semi-supervised Building Segmentation from Satellite Images: Using unlabelled data and pseudo-labelling to address data scarcity issues.
• Microscope Slide Cancer Cell Classification: An image classifier using edge detection to identify cancer cells in microscope slides.
• Pipeline Optimization for Image Augmentation: Enhanced a TensorFlow image pipeline, reducing processing time significantly.
• Scenery Classifier: Image classifier to categorize natural and urban landscapes.
• Land Use Satellite Classifier: A classifier to differentiate urban and rural areas adjacent to roads from satellite images.
• Roman Numeral Classification: Classifier for identifying handwritten Roman numerals.
• Microscope Slide Cell Segmentation: Image segmentation model to locate cell nuclei, aiding in cancer type prediction.

NLP Projects:

• Essay Grader: TensorFlow NLP model to predict grades for student essays.
• Movie Review Sentiment Classification: TensorFlow model for sentiment analysis on movie reviews.

🔍 Data Science/Classical ML/Data Preprocessing Projects

Data Science and Classical ML Projects Repo

This repository showcases my data science projects, including preprocessing and supervised learning. Highlights include:

Supervised Learning Projects:

• Energy Price Prediction - Automated Model Selection & Optimisation: Selecting the best-performing model and applying hyperparameter tuning for energy price prediction.
• Road Safety Regression Prediction: Using regression and ensemble learning for road safety scoring.
• Fraud Classification: Analyzing the Enron case using various classification algorithms.
• Employee Leave Prediction: Logistic regression model to predict employee absence patterns.

Preprocessing Projects:

• Data Preparation & Feature Analysis: Utilizing big data methodologies for efficient data processing and analysis.
• Energy Price Data Preprocessing: Aggregating and cleaning energy price data from various sources.
• Create Shuffled Train/Val Split for Image Datasets: Shuffling and combining image datasets into training and validation splits.

📊 Data Analysis Projects:

Data Analysis Projects Repo

This repo contains a number of Data Analysis projects I have explored, covering a range of topics and datasets, providing insights through exploratory analysis and visualization:

• A/B Testing and Regression: Statistical analysis related to E-Commerce Website click-through-rate.
• Data Visualization: Discovering patterns and displaying insights from bike rental usage data.
• Exploratory Data Analysis: Discovering trends and patterns in movie data.
• Data Wrangling and Analysis: Data scraping, cleaning and analysis of a certain Twitter Account.

🔍 SQL Examples:

SQL Examples: A demonstration of some SQL queries I wrote