## teaching_approach
How the teaching is structured
All three programmes follow the same general logic: describe the problem before offering solutions, use real artefacts (messy data, actual evaluation sets, code that runs) rather than illustrations, and give feedback that explains reasoning rather than just marking correctness or incorrectness.
Sessions are online and recorded. The development track adds fortnightly one-to-one office hours and a cohort channel, both of which are kept small enough to be practically useful. Each programme lists what it does not cover alongside what it does — so you can decide before signing up whether the scope matches what you need.
## programme_01 # data_handling_for_ai_projects
Data Handling for AI Projects
A five-week course on the part of AI work that consumes most of the time and gets taught least: getting data into a shape you can trust. Covers loading messy real-world files, schema validation, deduplication, leakage and how it sneaks into splits, class balance, labelling workflows and inter-annotator disagreement, dataset documentation, and versioning so that a result from March can still be reproduced in September.
# syllabus_topics
- Loading and inspecting genuinely untidy files
- Schema validation and deduplication strategies
- Data leakage — how it happens and how to detect it
- Class balance and labelling workflows
- Inter-annotator disagreement measurement
- Dataset documentation and reproducible versioning
# prerequisite
Python and basic pandas. No prior ML background required.
# this_course_does_not_cover
Model training, model architecture, deployment, inference optimisation, or evaluation methodology.
Ask About This Course## programme_02 # evening_seminar_evaluating_ai_systems
Evening Seminar: Evaluating AI Systems
A single three-hour evening seminar on the question most teams answer badly: is the new version better? Covers building a small evaluation set that reflects your actual traffic, choosing metrics that mean something, pairwise comparison, using models as graders and where that goes wrong, statistical significance on small samples, regression suites, and writing an evaluation report that a sceptical colleague would accept.
# seminar_topics
- Building an evaluation set from real traffic samples
- Selecting metrics that reflect what you actually care about
- Pairwise comparison methodology
- Model-as-grader approaches and failure modes
- Statistical significance on small evaluation samples
- Writing an evaluation report for a sceptical audience
# this_seminar_does_not_cover
Training, deployment, infrastructure, architecture decisions, or data preparation.
Ask About This Seminar
# included
- Live session + recording
- Slide deck for internal reuse
- Evaluation report template
- Reading list of primary sources
## programme_03 # ai_systems_development_track
AI Systems Development Track
A six-month, part-time track for developers building complete systems rather than isolated models. The syllabus moves from foundations through retrieval systems, tool-using agents and their failure modes, evaluation harnesses, cost and latency work, serving and monitoring, and finishes with a self-defined capstone that must run end to end and survive a review from someone who did not build it.
# syllabus_phases
- Foundations: vectorised computation, training loops, transformer end to end
- Retrieval systems and vector search
- Tool-using agents and their failure modes
- Evaluation harnesses and regression suites
- Cost, latency and serving infrastructure
- Self-defined capstone: end-to-end, independently reviewed
# prerequisite
Professional software experience, comfortable Python. Short technical conversation before enrolment.
# this_track_does_not_cover
Research methods, academic paper writing, fine-tuning at scale, or post-training alignment work.
Enquire About the Track## programme_comparison_matrix
Choosing between programmes
They are independent. You do not need to take them in sequence.
| Feature | Data Course | Eval Seminar | Dev Track |
|---|---|---|---|
| Duration | 5 weeks | 1 evening | 24 weeks |
| Time commitment | ~5 h/week | 3 hours total | 12–15 h/week |
| Live sessions | Weekly (2h) | One (3h) | Twice-weekly |
| Recording provided | |||
| Written feedback | 3 exercise sets | — | 4 projects + capstone |
| Office hours | — | — | Fortnightly 1:1 |
| Compute credit | — | — | |
| Cohort channel | — | ||
| Fee (RM) | 980 | 480 | 4,650 |
| Best for | Data engineers and ML engineers working with pipelines | Engineers, analysts, PMs evaluating AI output | Developers building complete AI systems |
## shared_standards_across_all_programmes
Personal data
Used only for course administration. Not shared with third parties for marketing.
Recording availability
All sessions recorded. Recordings shared with enrolled participants promptly after each session.
Syllabus changes
Changes between cohorts communicated in advance. Participants enrolled in a cohort receive what was described at the time of enrolment.
No outcomes claims
We make no claims about employment, salary, or career outcomes of any kind. These pages describe syllabuses and processes only.
## fee_table # Malaysian_Ringgit
| Programme | Fee (RM) | What is included |
|---|---|---|
| Data Handling for AI Projects | RM 980 | Live sessions, recordings, 3 exercise sets + written feedback, documentation template, cohort channel |
| Evening Seminar: Evaluating AI | RM 480 | Live session, recording, slide deck, evaluation report template, reading list |
| AI Systems Development Track | RM 4,650 | Twice-weekly sessions, recordings, 4 reviewed projects, capstone review, fortnightly office hours, cohort channel, compute credit, personal repository |
# prices are per participant per cohort · no additional tiers or hidden costs
## enquiry_cta
Have questions about a specific programme?
Use the enquiry form on the home page. We reply within two working days.
Send an Enquiry