Three courses, designed to stack
Each course addresses a distinct layer of applied deep learning. Start at the foundations, move to applied computer vision, then bring your own work to structured sessions.
← Back to HomeHow the learning is structured
Foundation first
Neural Network Fundamentals establishes the vocabulary — PyTorch primitives, training dynamics, regularization — that everything else builds on. Self-paced so students can move at the right speed.
Applied domain work
Computer Vision With Open Models applies the fundamentals to a concrete domain. The cohort format means weekly checkpoints, peer context, and facilitator review — not just watching videos.
Ongoing structured input
Office sessions extend the learning relationship for students doing independent project work. The quarterly cadence keeps engagement regular without requiring continuous formal enrollment.
Neural Network Fundamentals
A self-paced course covering the basic building blocks of neural networks — linear layers, activation functions, loss functions, training loops, and standard regularization approaches. The course works through small examples in PyTorch with full working notebooks. Aimed at students who can already write small Python programs and are comfortable reading mathematical notation at a moderate pace.
What the course covers
- Linear layers and matrix operations in PyTorch
- Activation functions: ReLU, sigmoid, softmax — when and why
- Loss functions and gradient descent mechanics
- Writing training loops from scratch — no high-level wrappers first
- Regularization: dropout, weight decay, early stopping
How it runs
- Python: can write small programs, read functions and loops
- Math: summation notation, partial derivatives at introductory level
- No prior machine learning experience required
- Maximum 30 students per cohort
- 10 weeks, weekly notebooks and exercises
- Final project submitted as written report
- Recommended: complete Fundamentals first
Computer Vision With Open Models
A ten-week course covering common computer-vision tasks built on top of open pretrained models — image classification, detection, segmentation, and basic representation learning. Students work through weekly notebooks, evaluation exercises, and a small final project documented as a written report. Each cohort is capped at thirty students.
Topics across the ten weeks
- Transfer learning: how and when pretrained weights work
- Image classification pipelines with torchvision models
- Object detection: YOLO-family and DETR-family approaches
- Semantic and instance segmentation fundamentals
- Representation learning and embedding spaces
Course structure
Quarterly Open Office Sessions
A quarterly pass for moderated open office sessions where students can bring their own projects and get structured feedback from facilitators. Each session is two hours and includes a short topical reading shared in advance. The sessions are intended for students who already have a course or project under way and want occasional structured input.
What to expect in a session
- Short topical reading circulated 3 days before the session
- First 20 minutes: discuss the reading as a group
- Remaining time: each student presents their project question
- Facilitator leads structured discussion and gives direct feedback
Who this is for: Students who have completed a Synaptiq course or are working on a serious ML project and want occasional, structured external input — not beginners looking for introductory help.
- One session in Jan–Mar quarter
- One session in Apr–Jun quarter
- One session in Jul–Sep quarter
- One session in Oct–Dec quarter
- Exact dates sent to pass holders 2 weeks in advance
Which course fits where you are
A quick reference to help you pick the right starting point.
| Feature | Fundamentals | Computer Vision ★ | Office Sessions |
|---|---|---|---|
| Price | ฿4,400 | ฿22,000 | ฿12,500/qtr |
| Format | Self-paced | Cohort · 10 weeks | Quarterly session |
| PyTorch notebooks | — | ||
| Facilitator feedback | — | ||
| Final project | — | — | |
| Best for | New to deep learning | Has Python + basic ML | Active project in progress |
Technical standards across all courses
PDPA compliant
Student data handled in accordance with Thailand's Personal Data Protection Act.
Tested on release
All notebooks run end-to-end on current PyTorch and ecosystem versions before each cohort opens.
Updated annually
No cohort receives curriculum older than 12 months. Post-cohort notes feed into the next update cycle.
1-day response
Enrollment and course questions answered within one business day at [email protected]
Clear pricing, no add-ons
- All course notebooks
- Module exercises
- No expiry on access
- Weekly notebooks + exercises
- Facilitator review of work
- Final project report
- Max 30 students
- 2-hour moderated session
- Topical pre-reading
- Bring your own project
Not sure which course to start with?
Send a brief message describing your current Python level and what you want to build. We'll point you in the right direction.
Ask a Question