Tensor Foundations & Autograd Craft
Build shape intuition through progressive exercises. Implement manual backprop on a two-layer network before touching nn.Module.
Six structured pathways from first tensor to production deployment. Each programme combines instructor-led labs, forge sessions, and an assessed capstone milestone at our King Street West campus.
Request programme detailsProgrammes build on each other. Prerequisites are enforced to protect cohort pace and peer learning quality. Cohorts begin monthly in 2026.
All programmes run in-person at our Toronto campus unless noted. Select a programme below for duration and entry requirements.
Build shape intuition through progressive exercises. Implement manual backprop on a two-layer network before touching nn.Module.
Bridge raw tensor skills and applied vision work through pandas pipelines, feature stores, and reproducible experiment tracking.
Implement convolutional layers from first principles, train on real datasets, and evaluate with metrics suited to imbalanced classes.
Tokenisation, embedding layers, and sequence-to-sequence architectures before students encounter transformer blocks in NCL-007.
Implement scaled dot-product attention from first principles. Fine-tune a small language model on domain-specific Canadian text corpora.
Design, train, and deploy an end-to-end model solving a real problem. Culminates in a presentation and written technical documentation.
Build shape intuition through progressive exercises. Implement manual backprop on a two-layer network before touching nn.Module. Graduate with a documented lab notebook proving you understand what autograd records.
Implement scaled dot-product attention from first principles. Trace positional encodings, multi-head attention, and layer normalisation. Fine-tune a small language model on domain-specific Canadian text corpora.
Individual students may split tuition into two or three monthly instalments at no interest. Payment plans available for all programmes. Corporate teams may request private cohort scheduling through our services team.
Every programme includes weekly code reviews, a mid-point skills checkpoint, and access to open forge hours on Saturdays. Weekend forge hours are included with every programme. Alumni retain access for six months after NCL-009 completion.
Not sure where to start? Book a free pathway advisory session. We map your current skills against programme prerequisites and recommend an entry point — even if that means self-study before NCL-001.
Each programme publishes a detailed syllabus on enrolment. Syllabi list weekly lab objectives, required pre-reading, assessment rubrics, and capstone milestones.
Request programme detailsEducation disclaimer: NeuralCraftLab programmes are non-credit professional training courses. They do not confer a degree or regulated professional designation. Programme content is updated regularly and may differ from prior cohorts.