Machine Learning & MLOps with Python Training Course

Who should attend

Analysts, Data Scientists, and software developers who want to get a practical introduction to Data Science and Machine Learning with Python.

Prerequisites

Delegates should ideally have some prior experience programming and / or of using statistical analysis techniques. Any experience with Python would be beneficial but we will give you "just enough" Python pointers to be able to create real-world workable solutions to data analysis and visualisation challenges.

Live, instructor-led online and on-site training

We appreciate that you need flexibility to fit in with new working situations - whether you're an individual, part of a distributed team, or simply have projects and deadlines to meet. 

Our remote training can take place online in a virtual classroom, with content split into modules to accommodate your scheduling challenges and meet your learning goals. Get in touch today to find out how we can help design a cost-effective, flexible training solution.

AI: What is it good for?

• Case studies
• When doesn't it work?
• Distilling the hype

How do machines learn?

• Supervised learning
• “Classic” ML vs Deep Learning
• Feature engineering
• Training
• Evaluation
• Describing the data vs predicting from it
• Data domains
• Tabular (unstructured) data, text, images, video, audio, multi-domain
• Unsupervised learning

Classic ML

• Supervised Learning Problems
• Classifier methods for predicting a label: Logistic regression, kNN, decision trees
• Regression methods for predicting a quantity: Linear regression, random forests
• State of the art ensemble methods: xgboost, catboost
• Evaluating and iterating on models
• Train/validation/test split, data leakage
• Underfitting/overfitting
• Feature selection, parameter tuning
• Model explainability, Shapley values
• Unsupervised Learning Problems
• Clustering: grouping similar items with k-Means
• Dimensionality Reduction with Principal Component Analysis

Fundamentals of Deep Learning with PyTorch

• Introduction to deep learning
• From the perceptron to the deep network
• How neural networks are trained
• When to use deep learning over classical ML
• Pretrained models, fine-tuning
• Building & training simple neural networks
• Overview of frameworks
• Layers, activation functions, loss functions
• Gradient descent: learning rate, batch size, epochs
• Overfitting & regularisation: dropout, L2
• Parameter tuning, auto-ML
• Training a neural network from scratch to recognise handwriting

MLOps

• Working with ML in production - from research to deployment
• Tooling
• Python Ecosystem for Data Science and Machine Learning
• Moving from notebooks to IDEs for local development
• Virtualenv, poetry, uv
• VSCode + Github Copilot, Cursor
• Cursor
• Experiment tracking with MLFlow
• Model deployment
• Docker containers
• Model endpoints:
• Custom endpoints with FastAPI
• Cloud endpoints - AWS Sagemaker, GCP, Azure
• Monitoring and alerting
• Model performance
• Data drift

Additional Topics for custom ML training

• Advanced methods
• Deep learning for images
• CNNs
• Vision Transformers (ViT)
• Generative modelling for creating new data
• Autoencoders
• GANs
• Style transfer
• Sequence/time series modelling
• Recurrent neural networks
• Transformers
• Text
• Transformers
• Fine tuning a BERT model
• Embeddings
• Reinforcement learning