Loading
Advanced Program
AI & Deep Learning
The mathematics, architectures and practice of modern neural networks.
- Length
- 12 or 18 months
- Level
- Advanced
- Weekly
- 10 to 15 hours
- Format
- Live cohort, Hybrid
- Modules
- 12 to 16
- Next intake
- To be announced
What you'll learn
A long-form program for people who want to understand deep learning from first principles and apply it across vision, language and generation. You implement core architectures yourself before using libraries. The 18-month option adds reinforcement learning, large-scale training and a research-style capstone.
By the end, you will be able to
- Derive and implement backpropagation and common optimisers
- Build CNNs, RNNs and transformers from components
- Train models stably with regularisation and good initialisation
- Apply self-supervised and transfer learning to limited data
- Implement and evaluate generative models including diffusion
- Read, reproduce and critique research papers
Curriculum
Switch between the 12- and 18-month options to see how the program expands. Every module includes a hands-on lab.
12 modules, 48 weeks, about 480 to 720 study hours
Core modulesCapstone
Topics
- Vectors, matrices and gradients
- Chain rule
- Probability and likelihood
- Information theory basics
Lab
Implement gradient descent from scratch
Estimated effort
40 to 60 hours over 4 weeks
Skills you'll build
- Deep learning theoryAdvanced
- TransformersAdvanced
- Computer visionProficient
- Generative modelsProficient
- Reinforcement learningProficient
- Research methodsProficient
Tools and technologies
- Python
- PyTorch
- NumPy
- Hugging Face
- Weights & Biases
- CUDA
- Jupyter
Product names are trademarks of their owners, listed only to describe what we teach. No endorsement is implied.
Portfolio projects
Representative projects from the curriculum. Your capstone is scoped with a mentor around a problem you choose.
A small GPT from scratch
Implement and train a decoder-only transformer and study how it scales.
PyTorch, Transformers
Label-efficient image classifier
Use self-supervised pretraining to reach strong accuracy with 5% of labels.
Contrastive learning, CNNs
18-month option
Paper reproduction
Reproduce a published result, run ablations and report what holds up.
Research methods
How it's delivered
- Live cohort
- Scheduled live classes each week with your cohort, recorded for different time zones, plus self-paced lessons and labs.
- Hybrid
- Self-paced lessons with fortnightly live workshops and mentor sessions.
Instructors
Instructor and mentor profiles for this program are published once each person's background has been verified. We don't show placeholder faces or borrowed credentials.
Admissions
- 1Submit an online application
- 2Short technical conversation
- 3Decision by email
- 4Confirm your place and start onboarding
Tuition
- 12-month optionPublished before intake
- 18-month optionPublished before intake
Payment options vary by country and are shown at checkout.
Tuition and payment optionsCertificate
Complete every required module, pass the assessments and the capstone review to earn the Advanced Program.
Each certificate has a unique ID and QR code that anyone can check on our verification page.
This is a professional credential issued by Zyvora Academy, not an academic degree.
Where it can lead
Roles this program builds toward. Outcomes depend on your background, effort and local job market; we don't guarantee employment.
Machine Learning Engineer
Turns data into models and models into dependable services.
Role detailsComputer Vision Engineer
Builds systems that interpret images and video.
Role details
Program questions
Syllabus
Get the full syllabus
We'll email the detailed syllabus, and the fee sheet as soon as tuition for the next intake is published.
Advisor
Speak to an advisor
Ask about fit, prerequisites or which length suits you. An advisor replies by email to arrange a call.
