LAUNCHPOINT / THE LEARNING + TRAINING CAMPUS
Learn with direction.
Train with purpose.
A complete view of the proposed skill architecture: 12 common-core courses and 30 specialist plans across eight fields. Every code opens the subjects, practical work and evidence expected.
Engineering foundations: Python, Linux & Git
Shell navigation · Functions and modules · Working tree, index and commits
Mathematics for learning systems
Vectors and matrices · Derivatives · Conditional probability
Data science: SQL, EDA & statistical reasoning
NumPy arrays · Missingness · Sampling
Machine learning: baselines to reliable decisions
Linear and logistic regression · Random forests · K-means
Deep learning: DNNs, PyTorch, CNNs & sequences
Dense neural networks · Tensors and autograd · Convolutions
Language engineering: text to transformers
Unicode and tokenisation · Bag-of-words · Queries, keys and values
Generative AI, RAG & bounded agentic workflows
Pretraining versus inference · Chunking · Tool contracts
Software delivery for AI: APIs, tests, Docker & CI
HTTP and JSON · Unit and integration tests · Images and containers
MLOps: versioning, experiments, pipelines & monitoring
Git versus data bytes · MLflow parameters, metrics and artefacts · Explicit dependencies
Scientific modelling, digital twins & research practice
Governing equations · System identification · State estimation
Engineering communication & career evidence
Issue writing · Architecture diagrams · Repository hygiene
Industry capstone & individual technical defence
Problem charter · Baseline · Failure analysis
Learn together.
Make it work.
The common core gives everyone a shared engineering foundation. A focused pathway gives you the time to deepen a real piece of work.
Choose a specialist direction ↗Actual studio photograph · earlier programme archiveData engineering, ETL & quality contracts
Relational schemas · Ingestion · Schema
Large-scale data processing & Spark
Partitioning · Joins · Feature tables
Optimisation, uncertainty & robust decisions
Convexity · Aleatoric and epistemic uncertainty · Distribution shift
Classical AI: search, reasoning & planning
State spaces · Propositional logic · Goals and actions
Feature engineering & feature systems
Encoding · Ablations · Entity keys
Ensembles, XGBoost & interpretable decisions
Residual fitting · Validation design · Feature attribution
Causal inference & experimentation
Association · Randomisation · Adjustment
Time-series forecasting & anomaly detection
Seasonality · Persistence · Statistical forecasts
Recommender systems & ranking
Explicit and implicit feedback · Popularity baseline · Pointwise and pairwise learning
Graph machine learning & knowledge graphs
Nodes and edges · Message passing · Triples
Computer vision: detection & segmentation
Annotation rules · Bounding boxes · Pixel labels
3 D vision & spatial intelligence
Camera models · Depth maps · Registration
Advanced neural architectures & efficient training
Residual connections · Initialisation · Mixed precision
Generative modelling: VAEs, GANs & diffusion
Latent representations · Generator and discriminator · Noise schedules
Production RAG, search & evaluation
Lexical search · Chunk strategies · Reranking
Agent engineering, tool use & evaluation
State machines · Typed tool inputs · Approval checkpoints
Audio, speech & multimodal learning
Sampling · Recognition · Image-text representations
Language-model adaptation & efficient fine-tuning
Prompting · Instruction data · Low-rank adaptation
Kubernetes & platform engineering for ML
Pods · Requests and limits · Batch jobs
Cloud architecture & model-serving systems
Batch · Authentication · Batching
Distributed training, GPUs & performance engineering
CPU and GPU roles · Data and model parallelism · Data loading
Edge AI, IoT & efficient inference
Latency · Quantisation · Sensor contracts
Scientific computing, numerical methods & twin engineering
ODEs · Parameter estimation · Filtering concepts
Research methods, reproducibility & publication craft
Question · Baselines · Methods
Responsible AI, privacy & AI-system security
Stakeholders · Minimisation · Prompt injection
Reinforcement learning & sequential decisions
States · Value functions · Reward specification
Deep reinforcement learning: DQN & policy gradients
DQN · Policy gradients · Seed variation
ROS 2: nodes, messages & robot software
Nodes · Topics · Workspaces
Robot perception, localisation & SLAM
Transforms · Camera and range data · Odometry
Robot planning, control & sim-to-real boundaries
Configuration spaces · Feedback · Planner-controller interfaces
No matching course plan.
Try a broader subject or clear the filters.
AVAILABLE TO STUDY NOW
Start with a
working experiment.
Four original reading/practice modules, executed notebooks, knowledge checks and browser-local notes. These are foundation modules, not the entire 42-course catalogue.
MINDFORGEAI / LABA defensible regression baseline
Turn a table into a model, then test whether it earns its complexity.
MINDFORGEAI / LABClassification beyond accuracy
Use neighbours, scaling and a confusion matrix to understand a decision model.
MINDFORGEAI / LABBuild an evidence-grounded retrieval baseline
Find a source, inspect its relevance and recognise what a similarity score cannot tell you.
MINDFORGEAI / LABFrom a differential equation to a learned rate
Fit a dynamical model to observations, then test what the model can responsibly predict.