Level 3 of 3
Advanced Level
Prerequisites
Deep Learning Intermediate level, or equivalent: structured PyTorch training code, modern CNNs, data set building and rigorous evaluation.
Duration
12 weeks · 140 guided hours · 6 modules. Taught live by an instructor — online or in person.
By the end of this level you can
- Explain and fine-tune transformer models for language and vision tasks
- Describe how generative models work and assess their risks
- Build detection and segmentation models and evaluate them correctly
- Deploy deep learning models efficiently, including offline on phones
- Develop deep learning systems responsibly and document them
Modules
- 1Module 1 · 24 h · 4 lessonsAttention and Transformers
- 2Module 2 · 20 h · 4 lessonsGenerative Models
- 3Module 3 · 22 h · 4 lessonsComputer Vision Beyond Classification
- 4Module 4 · 22 h · 4 lessonsDeploying Deep Learning Models
- 5Module 5 · 22 h · 4 lessonsResponsible Deep Learning
- 6Module 6 · 30 h · 5 lessonsDeep Learning Capstone
Advanced level project
Offline Field Assistant Model
Design, train, compress and evaluate a deep learning model that runs offline on a mid-range Android phone for a field task — for example detecting crop pests in photos or classifying short voice or text reports in a local context — with documentation of its limits and risks.
Deliverables
- A trained model (transformer-based or CNN-based) with a reproducible training repository
- A compressed, exported model with measured on-device size and speed
- An evaluation including field-like test data and group or condition breakdowns
- A model card and a risk assessment, plus a demonstration
Assessed on
- The model meets agreed accuracy, size and latency targets on the target device
- Evaluation reflects field conditions and reports failure modes honestly
- Data use respects consent, privacy and licences
- Documentation lets another team retrain, deploy and monitor the model
All levels are taught live by an instructor — online or in person — see registration options.