Bootcamps
Coming soon · the subjects below are what they will draw on
Subjects
What it is
Image formation, features, detection, segmentation, and convolutional models.
Where it sits
School of Artificial Intelligence · Upper level
Crunch Vision — Computer Vision
You get
Images as data, filtering and transforms, classical features, CNNs and transfer learning, object detection, segmentation, motion and tracking, and Vision Transformers.
What it is
Backpropagation, convolutional and transformer architectures, training dynamics, and transfer learning.
Where it sits
School of Artificial Intelligence · Upper level
Crunch Nets — Deep Learning & Neural Networks
You get
Deep learning built from the gradient up: an autograd engine and multi-layer network written by hand in NumPy, then optimizers, initialisation, normalisation and regularisation; CNNs, RNNs and LSTMs, self-attention and a Transformer block from scratch, embeddings and autoencoders, VAEs and GANs, and the trained model exported, quantised and served.
What it is
Diffusion and transformer generation, fine-tuning, retrieval augmentation, and evaluating generated output.
Where it sits
School of Artificial Intelligence · Upper level
You get
Generation reasoned about from the distribution up: latent-variable models, GANs, diffusion, steering and control, language generation, multimodal work, and evaluation that is honest about what came out.
Crunch Agents — AI Agent Systems Engineering
You get
The systems half — serving, retrieval, tool use, multi-agent orchestration, evaluation and production operations.
Still to add
Forty-eight of the exercise files — every week indexes three exercises and ships one, so the second and third are listed in all twenty-four weeks and not yet written.
What it is
Feature stores, model registries, deployment, drift monitoring, and retraining in production.
Where it sits
School of Artificial Intelligence · Upper level
Crunch MLOps — Production Machine Learning
You get
The production side of machine learning: packaging and serving models, registries and versioning, pipelines, monitoring for drift, and retraining without breaking what is live.
What it is
Tokenization, embeddings, sequence models, transformers, and language tasks end to end.
Where it sits
School of Artificial Intelligence · Upper level
Crunch Language — Natural Language Processing & Large Language Models
You get
Text turned into tensors, classical baselines first, embeddings trained and read, sequence models, attention and a Transformer implemented, encoder and decoder models, fine-tuning and transfer, grounding an LLM, and evaluation that does not flatter the model.
What it is
Markov decision processes, value and policy methods, exploration, and policy-gradient algorithms.
Where it sits
School of Artificial Intelligence · Upper level
Crunch RL — Reinforcement Learning
You get
Decision problems formalised, exploration versus exploitation, MDPs solved exactly, learning from experience, value-based and policy-based agents, continuous control, model-based planning, and alignment from human feedback.