What you should be able to do
Design neural architectures
Train with stable objectives
Diagnose under/overfitting
Deploy inference services
Every module is now a full lesson.
MLP
Build feed-forward neural networks and understand optimization behavior.
CNN
Learn local spatial patterns with shared convolutional filters.
Sequence models
Build a working mental model of Sequence models then connect it to implementation evaluation and production trade-offs.
Attention
Let a model dynamically weight which context elements matter for each representation.
Transformers
Build sequence models around attention, residual paths and position information.
Optimization
Build a working mental model of Optimization then connect it to implementation evaluation and production trade-offs.
Regularization
Build a working mental model of Regularization then connect it to implementation evaluation and production trade-offs.
Projects that prove the skill
Intent router
Use a baseline first then instrument the result and document failure cases.
design with cortex →Text classifier
Use a baseline first then instrument the result and document failure cases.
design with cortex →Vision baseline
Use a baseline first then instrument the result and document failure cases.
design with cortex →How progress is judged
Training curves
Evidence should come from code, results, explanation quality and the ability to identify when an approach should not be used.
Ablation
Evidence should come from code, results, explanation quality and the ability to identify when an approach should not be used.
Latency and memory
Evidence should come from code, results, explanation quality and the ability to identify when an approach should not be used.
Turn the track into a personal plan.
Cortex can break this path into daily sessions and adjust based on completed lessons, interview results and your saved learning goals.
Ask Cortex to plan it →