About Corvid Labs
Built to teach AI the way engineers actually need to understand it
We are a small school. That is a deliberate choice, not a constraint we are working around.
Back to HomeOur Story
How Corvid Labs came to exist
Corvid Labs grew out of a frustration shared by several engineers working in Malaysia's growing data and AI sector: most available training either taught library calls without explaining what was happening underneath, or covered theory in ways that never connected to working code. The gap between the two left practitioners who could run a training script but could not reason about why it was misbehaving, or who could derive a loss function but did not know how to implement the backward pass.
The school was established in George Town, Penang, in part because the city has a long tradition of technical education and in part because a significant number of the engineers and researchers we wanted to teach were already based in the northern corridor. Operating in English from the outset made it possible to draw learners from across the region and to use the same literature and notation that appears in the papers being discussed.
We keep cohorts small. This is not cost-efficient and we know it. The reason is that the kind of feedback that actually changes how someone reasons about a problem takes time to write and requires reading their specific work, not a generic submission. Large cohorts make that kind of feedback impossible, so we do not run them.
Mission
To produce engineers and researchers who understand the mechanisms behind the tools they use — people who can read a paper, identify the assumptions, and build from them.
What we do not do
We do not teach prompt engineering, no-code AI tools, or any course structured around a single library version. These have their place but they are not what we are building.
Where we are
Our offices are at 33 Jalan Burma, George Town, Penang. Sessions are delivered online; the address is where we operate from and where occasional in-person events are held.
The People
Who teaches and runs the programmes
Ahmad Zulkifli
Lead Instructor — Mathematics & Architectures
Teaches the mathematics and transformer architecture cohorts. Previously spent several years in applied research at a compute laboratory in Kuala Lumpur, focused on efficient training at limited scale.
Nur Liyana
Research Programme Director
Runs the research programme and supervises participants from literature survey through to their assessed talk. Her own research background is in statistical learning theory and evaluation methodology.
Rajan Krishnamurthy
Curriculum and Problem Set Design
Designs the problem sets and derivation clinics across all three programmes. Spent eight years writing educational material for engineering degree programmes before joining Corvid Labs at its founding.
Standards
How we maintain quality across cohorts
Every submission is read
Problem sets are marked by an instructor, not a script. Comments address the specific approach taken, not a model answer. This takes more time; it is also the point.
Curriculum reviewed each cohort
Materials are updated between cohorts when feedback from participants identifies areas that moved too quickly, relied on unstated assumptions, or where the field has shifted enough to matter.
Data handling and privacy
Participant data is stored only as long as required for the programme and communications related to it. Submission materials are not used for any purpose other than feedback. See our Privacy Policy for detail.
Sample sessions before enrolment
Full recorded sessions are available before any payment is made. Prospective learners are encouraged to watch at least one complete session from the course they are considering.
Prerequisite honesty
Course prerequisites are written to be accurate rather than permissive. We would rather turn away an application from someone who is not ready than have them pay for a course and struggle from the start.
Post-cohort feedback loop
At the end of each cohort, every participant is asked a short set of specific questions about what worked and what did not. The answers inform the next version of the material directly.
Corvid Labs · George Town · Penang
Corvid Labs operates as an online school from George Town, Penang, offering courses in the mathematical and computational foundations of AI development. The courses are taught in English and are open to learners across Malaysia and the wider region. Sessions are held online with fixed schedules; all materials, replays and problem sets are accessible through the cohort platform.
The three current programmes — mathematics for machine learning, transformer architectures, and research practice — cover distinct but related areas. The mathematics programme addresses the core linear algebra, calculus and probability that appear repeatedly in AI literature and that are often assumed rather than taught. The transformer architecture programme covers the full implementation of attention-based models from the ground up, including the training and inference considerations that matter in practice. The research programme works at the level of methodology and writing, preparing participants to contribute to the field as practitioners or researchers.
Each programme is taught in cohorts with a defined start date, a fixed number of weekly sessions, and a clear end point. Mentor calls and submission feedback are included in the course fee. Compute credits for the architectures and research programmes cover the cloud resources needed for the assignments and projects within each course.
Interested in joining a cohort?
Send us a note about the programme you have in mind and your current background. We will let you know whether the fit looks right and when the next cohort opens.
Contact Corvid Labs