George Town · Penang · Malaysia
The mathematics and architecture
behind AI — taught with rigour
Cohort-based courses that take you from first principles to working code. Every derivation runs alongside Python. Every cohort has a fixed schedule and a real end date.
Current Cohorts
Three programmes. One discipline.
Each course builds on a specific slice of the AI development stack. They can be taken in order or selected according to where you currently stand.
Mathematics for Machine Learning
Eight weeks covering the mathematics that modelling work actually uses — vectors, matrices, eigendecomposition, gradients, probability and information measures. Every idea is implemented in Python the same session it is derived.
- 8 live sessions with replays
- Weekly problem sets, worked solutions
- Fortnightly derivation clinic
- One mentor call included
Transformer Architectures in Depth
Twenty-four weeks building sequence models from scratch. Tokenisation, embeddings, self-attention, multi-head attention, full training loops, then fine-tuning, quantisation, inference optimisation and paper reproduction at small scale.
- 24 live sessions and compute credits
- Weekly code review
- Paper reproduction project
- Four mentor calls, architecture defence
Research Practice Programme
Thirty-two weeks for learners moving toward research or applied research roles. Experimental design, ablation discipline, writing for peer review, poster preparation and a full research question carried from survey to presented talk.
- Supervisor assigned throughout
- Fortnightly writing clinics
- Full paper draft, two review rounds
- Presented talk with assessment
Why Corvid Labs
What makes the teaching here different
Derivations before APIs
Every concept is built from first principles before any library is opened. The goal is understanding the mechanism, not just calling the function.
Code alongside every derivation
Python implementations follow each mathematical result in the same session. Theory and practice are kept in step throughout the course.
Fixed schedules, real deadlines
Cohorts start on a set date and end on a set date. Problem sets are due weekly. This structure is deliberate — most self-study stalls without it.
Small cohorts
Cohort sizes are kept small enough that instructors can read every submission and respond with comments that are specific to the work, not generic.
Misconception panels in every lesson
Each teaching area includes a panel listing the mistakes learners most often make at that point, with brief corrections. Errors are treated as data, not failures.
Sample sessions, not marketing copy
Full recorded sessions are available before enrolment so prospective learners can judge the teaching quality for themselves rather than take our word for it.
Ready to start?
The next cohort opens shortly.
Places are limited by design.
Send us a short message with the course you are considering and where you currently stand. We will follow up within two working days with cohort dates and prerequisites.
Questions
Frequently asked
Who are these courses for?
Do I need to attend sessions live?
How much time should I set aside each week?
What is the fee structure and how do I pay?
Is there a completion record or certificate?
Are the courses taught in English?
How do I know if my current level is right for a course?
Find Us
Our Location
33 Jalan Burma, 10050 George Town, Pulau Pinang, Malaysia
Contact
Write to us about a cohort
Contact Details
Phone
+60 4-899 3742Address
33 Jalan Burma
10050 George Town
Pulau Pinang, Malaysia
Office Hours
Monday – Friday: 9:00 am – 6:00 pm
Saturday: 10:00 am – 1:00 pm
Sunday: Closed
We reply to most enquiries within two working days. For urgent matters, please call during office hours.