Three Programmes.
Three Points in the Stack.
Each programme sits at a defined layer in the AI development system. Choose the layer that matches your current position and prior knowledge.
Back to HomeHow the Stack Approach Works
AI development systems have structure. Language models sit on top of training data, hardware and optimisation decisions; fine-tuned models sit on top of base models; deployed systems sit on top of inference infrastructure. Understanding each layer means understanding its inputs, outputs and the constraints it passes upward.
Bayang Labs programmes are designed around this structure. Each session identifies which layer it is addressing, what comes from below, and what it passes up. Learners who go through multiple programmes build a connected map rather than a collection of isolated techniques. Learners taking a single programme get a clear location in the broader system, not just a set of skills without context.
Assumed Knowledge — L01
You are comfortable using software at work and do not need any programming background.
Introduction to Language Models for Non-Engineers
A four-week evening programme for product managers, analysts, designers and operations staff who work alongside engineering teams. Covers what language models do and do not do, tokens and context, prompting as specification writing, evaluating output quality, cost drivers, and where these systems tend to fail. Includes a weekly live session, short practical exercises, and a written glossary in English and Malay.
- What language models do and do not do
- Tokens, context windows and their practical consequences
- Prompting as specification writing
- Evaluating output quality and identifying failure modes
- Cost drivers and when these systems are the wrong tool
- Bilingual English–Malay glossary
Programme Steps
- 01 Week 1 — System overview: what a language model is and is not
- 02 Week 2 — Inputs: tokens, context, prompts as specifications
- 03 Week 3 — Outputs: evaluating quality and recognising failure
- 04 Week 4 — Deployment: cost, latency, appropriate use cases
RM 780
Duration: 4 weeks (evening sessions)
Enquire About L01
Assumed Knowledge — L02
You write Python comfortably and have completed at least one deep learning course covering training loops and gradient descent.
Fine-Tuning and Adaptation Programme
A ten-week programme covering parameter-efficient adaptation, dataset construction for supervised fine-tuning, preference data, evaluation before and after adaptation, deployment of adapted open-weight models, and honest assessment of when adaptation is the wrong tool. Includes GPU credits, three individual mentor sessions, weekly code review, and a capstone project.
- Parameter-efficient adaptation methods (LoRA, QLoRA)
- Dataset construction for supervised fine-tuning
- Preference data and RLHF concepts
- Evaluation metrics before and after adaptation
- Deployment of adapted open-weight models
- GPU credits + weekly code review included
Programme Steps
- 01 Weeks 1–2 — Base model selection and parameter-efficient methods
- 02 Weeks 3–4 — Dataset construction and data quality
- 03 Weeks 5–6 — Training runs, monitoring and debugging
- 04 Weeks 7–8 — Evaluation and preference alignment
- 05 Weeks 9–10 — Capstone project and deployment
RM 3,640
Duration: 10 weeks · Includes GPU credits and 3 mentor sessions
Enquire About L02Assumed Knowledge — L03
You have completed technical AI or ML coursework and have project work you want to present clearly to reviewers.
Portfolio and Interview Preparation Support
A six-week supplementary programme for learners who have finished technical coursework and want their work presented clearly. Covers repository structure and documentation, writing up a project so a reviewer understands the decisions, preparing a short technical talk, whiteboard practice on model and systems questions, and two recorded mock interviews with written feedback. Offers preparation and practice only — no employment arrangements or introductions of any kind.
- Repository structure and README writing
- Writing up decisions so reviewers follow the reasoning
- Preparing a 10-minute technical talk
- Whiteboard practice on model and systems questions
- Two recorded mock interviews with written debrief notes
Programme Steps
- 01 Weeks 1–2 — Repository and documentation structure
- 02 Week 3 — Writing project write-ups that explain decisions
- 03 Week 4 — Preparing and practising a technical talk
- 04 Week 5 — Whiteboard practice, model and systems questions
- 05 Week 6 — Two mock interviews + written feedback
RM 1,340
Duration: 6 weeks · Includes 2 recorded mock interviews with written feedback
Enquire About L03
Which Programme Fits You
| Feature | L01 Intro | L02 Fine-Tuning | L03 Portfolio |
|---|---|---|---|
| Programming required | None | Python + DL | Technical background |
| Duration | 4 weeks | 10 weeks | 6 weeks |
| GPU credits | — | — | |
| Mentor sessions | — | 3 individual | — |
| Mock interviews | — | — | 2 recorded |
| Bilingual glossary | — | — | |
| Price | RM 780 | RM 3,640 | RM 1,340 |
Best for L01 if…
You use AI tools at work and want to understand what is actually happening, or you collaborate with engineers and want to write better requirements.
Best for L02 if…
You have training experience and want to go deeper on adaptation — choosing when to fine-tune, how to build the data, and how to evaluate properly.
Best for L03 if…
You have projects already built and want to present them in a way that makes the decisions visible and the scope clear to someone reviewing your work.
Technical and Delivery Standards
Data Privacy
Learner submissions and personal data are handled under our Privacy Policy. No sharing with third parties for purposes outside programme delivery.
Current Materials
Content is reviewed before each cohort to remove outdated tooling references and reflect significant changes in the field.
Small Cohorts
Programmes are capped to maintain meaningful participation. Size per cohort is confirmed in the programme description each run.
Scope Transparency
Each programme states what it does not cover. The portfolio programme explicitly offers preparation and practice only, with no placement or introduction services attached.
Current Tooling
The fine-tuning programme uses open-weight models and parameter-efficient methods actively in use in production environments, not historical examples.
Written Records
Feedback at every level is written and handed to the learner. Code review notes, mentor session summaries and mock interview debriefs all become the learner's documents to keep.
Know Your Layer. Enquire Today.
Contact us with your background and the layer you are aiming for. We will confirm availability and walk through next steps.
Get in Touch