Three stalls. Three distinct problems. Read the signs before you step in.
Full details on each course: what it covers, what it does not, how it is delivered, what you leave with, and what it costs.
← Back to HomeHow the courses are structured
Each course is a self-contained programme with a fixed number of weeks, a stated weekly hour range, defined prerequisites, and a clear list of what the course covers. The scope does not expand because you ask nicely, and it does not shrink because the material is harder than expected.
All courses include instructor-written code review on every assignment. None of them use automated scoring as a substitute. Live contact with the instructor — whether weekly sessions, office hours, or design reviews — is scheduled in advance and is part of the programme, not an optional extra.
Pricing is per individual for the two standalone courses. The Enterprise Programme is priced per engineer within the cohort and is not available to individuals enrolling separately. All fees are in Malaysian Ringgit.
Practical Computer Vision Engineering
This course covers the engineering work involved in training and deploying vision models for ordinary industrial problems. The focus is on the full pipeline: labelling practice, dataset curation, augmentation strategies (including those that hurt performance and why), model training, evaluation under distribution shift, and deployment to an edge device.
The course is written for engineers with Python experience who work near cameras, sensors or manufacturing lines. It does not begin from "what is a neural network." It begins from "you have a labelling budget, a distribution shift problem, and a device with limited compute."
Who this is not for: anyone building systems that identify individuals, learners without working Python experience, or anyone expecting a no-code workflow.
What is included
- Weekly live sessions with the instructor
- Ten assignments with written code review on each
- Labelling quality exercise using a public dataset with known issues
- Deployment exercise to an edge device
- Written assessment record on completion
How the course runs
- 1Weeks 1–3: Dataset construction, labelling practice, quality review
- 2Weeks 4–6: Model selection, training, augmentation analysis
- 3Weeks 7–9: Evaluation under distribution shift, failure analysis
- 4Weeks 10–11: Edge deployment, final assessment
Retrieval Systems and Model Evaluation
This course addresses retrieval-augmented systems and — more centrally — the problem of measuring whether they work. It covers embedding choice, chunking strategy, hybrid search, reranking, and the construction of evaluation sets that are not merely flattering. A significant portion of the course is dedicated to adversarial evaluation and annotator agreement.
The course is suited to engineers who have shipped at least one AI feature to a production environment and found it hard to say with confidence whether it improved anything. The course teaches how to measure that confidently, and leaves the learner with a reusable evaluation harness.
Who this is not for: learners who have not yet deployed an AI feature, anyone looking for a RAG tutorial without evaluation rigour, those unwilling to write the evaluation harness from scratch.
What is included
- Cluster access for the duration of the course
- Twelve assignments with written code review on each
- Full evaluation harness written by the learner and kept afterwards
- Adversarial evaluation exercise
- Session on annotator agreement and cost
- Weekly office hours with the instructor
- Public write-up produced as a course deliverable
How the course runs
- 1Weeks 1–3: Retrieval fundamentals, embedding selection, chunking
- 2Weeks 4–6: Hybrid search, reranking, pipeline construction
- 3Weeks 7–10: Evaluation set construction, annotator agreement, adversarial sets
- 4Weeks 11–14: Harness completion, public write-up, final review
Enterprise AI Enablement Programme
A cohort programme for Malaysian technology teams, banks' engineering divisions and manufacturers with in-house data. Eight to sixteen engineers from the client organisation train on an internal problem defined during the scoping study. The programme does not run on a generic curriculum — the problem, the data and the regulatory context are the client's own.
The curriculum includes a module on data governance, provenance and consent built on published Malaysian requirements. All code produced during the programme is owned by the client organisation outright. The programme closes with an internal technical presentation and a written capability assessment for each participant.
Who this is not for: individuals enrolling outside an employer-sponsored cohort, organisations without an in-house engineering team, teams with fewer than eight eligible engineers.
What is included
- Scoping study with the client's engineering leadership
- Tailored curriculum built around the client's chosen problem
- Cluster and inference budget for the cohort
- Weekly mentored design reviews
- Shared internal codebase (client owns outright)
- Data governance, provenance and consent module
- Internal technical presentation at close
- Written capability assessment per participant
How the programme runs
- 1Pre-start: Scoping study, problem definition, curriculum draft
- 2Weeks 1–6: Foundations, data audit, problem framing
- 3Weeks 7–14: Core engineering work, design reviews, codebase development
- 4Weeks 15–20: Data governance module, integration, assessment
- 5Weeks 21–22: Internal presentation, capability assessments issued
Choosing the right stall
Use this table to narrow down which course fits your situation. The prerequisites column is the most important one to check first.
| Feature | Computer Vision RM 475 |
Retrieval Systems RM 1,640 |
Enterprise Programme RM 4,390/engineer |
|---|---|---|---|
| Prerequisite | Python + numerical libs | Shipped an AI feature | Team of 8–16 engineers |
| Duration | 11 weeks | 14 weeks | 22 weeks |
| Weekly hours | 8–10 | 10–12 | 12–15 |
| Compute included | Edge device exercise | Cluster access | Cluster + inference budget |
| Live instructor contact | Weekly sessions | Weekly office hours | Weekly design reviews |
| Deliverable you keep | Written assessment record | Evaluation harness + public write-up | Client-owned codebase |
| Best for | Engineers entering vision work near cameras or sensors | Engineers who can't yet measure what they built | Teams building on an internal problem under Malaysian governance |
Standards applied across all courses
Learner data
Submission data and assessment records stored for the course duration plus 12 months, then deleted. Not sold or shared with third parties.
Structured feedback
Mid-course and end-of-course reviews collected from every cohort. Changes to materials resulting from feedback are logged in a per-course changelog.
Scope in writing
Prerequisites, course scope, exclusions and "who this is not for" published before enrolment opens. Not adjusted after payment is received.
Invoice-based payment
Corporate invoice available for individuals and enterprise cohorts. No hidden fees inside the course. The stated fee is the full fee.
Code review practice
Every assignment receives a written review from an instructor who read the specific submission. Not automated scoring, not peer review.
No outcome representations
We make no representations about jobs, salaries or career outcomes. We teach engineering practice; we do not represent what the market will do with it.
Course fees
- 11 weeks
- 10 assignments + code review
- Weekly live sessions
- Written assessment record
- 14 weeks
- 12 assignments + code review
- Cluster access included
- Evaluation harness + write-up
- 22 weeks
- Scoping study included
- Cluster + inference budget
- Client-owned codebase
Still deciding which course fits?
Send us a brief description of your background and what you want to build. We will point you to the right stall — or tell you if none of them fit.
Contact Us