Neuralaya
Developer working with code and ML tools
[solutions, 3, courses]

Three courses. Scope and pricing stated plainly.

A four-week short course, a two-day intensive, and a five-month part-time programme — each with its own syllabus, workload, prerequisites and price.

Back to Home
[approach, method, overview]

How our courses are structured

Prerequisites layer

Every course starts with what you need to know already. This is the first layer. If the prerequisite doesn't match your situation, we'd rather you know before joining than after.

Workload layer

Session hours and homework hours are stated as integers alongside the course name. You can check them against your calendar before reading any description.

Scope layer

Each course includes a "not covered here" section. This is how you can tell whether a course fits what you're trying to learn, rather than discovering a gap halfway through.

Language model API request diagram
[course_1, batch, tokens]

Working with Language Model APIs

RM 890 · 4 weeks · 2 hrs/session + ~3 hrs homework/week

Prerequisite

You write JavaScript or Python professionally. No prior ML background needed.

A four-week short course for developers who need to build reliable features on top of hosted language models rather than train their own. The course covers prompt structure and why small wording changes move outputs, structured output and schema validation, tool calling, streaming, token accounting and cost control, caching, retries and graceful degradation when a provider is slow.

The section most tutorials skip: evaluation. How to tell whether a change actually improved anything rather than felt better. Every example is written in plain functions rather than class hierarchies, matching the codebase style most teams here work in.

What's covered

  • Prompt structure and output determinism
  • Structured output and schema validation
  • Tool calling and streaming
  • Token accounting, caching, retries
  • Evaluation: measuring improvement objectively

Not covered here

Training or fine-tuning models. Deployment infrastructure. Reinforcement learning. Large-scale data pipelines.

Included

Live sessions, recordings, a reference repository, an evaluation harness you keep, and written feedback on two exercises.

Process steps

1Send enquiry and confirm enrolment
2Receive course materials and repository access
3Attend four live sessions over four weeks
4Complete two exercises; receive written feedback
Enquire About This Course
[course_2, adapter, weights]

Weekend Intensive: Fine-Tuning Small Models

RM 1,340 · 2 days · 6 hrs/day · max 16 participants

Prerequisite

Comfortable with PyTorch basics. You should be able to run a training loop, not just read one.

A two-day intensive on adapting small open-weight models to a narrow task on modest hardware. Day one covers dataset construction — and why it dominates everything else — formatting, deduplication, splits, and honest baselines. Day two covers parameter-efficient adaptation, hyperparameter choices that matter versus those that do not, evaluation against the baseline, and knowing when the answer is that fine-tuning was the wrong tool.

You leave with a trained adapter, an evaluation report, and the notebooks. Narrow by design: this course does not cover large-scale training, reinforcement learning from human feedback, or deployment.

What's covered

  • Dataset construction, formatting, deduplication
  • Honest baselines before any adaptation
  • Parameter-efficient adaptation (LoRA / QLoRA)
  • Hyperparameter choices that matter
  • Evaluation against baseline; go/no-go decision

Not covered here

Large-scale distributed training. RLHF. Deployment and serving. API wrappers around the adapted model.

Included

Sessions, recordings, notebooks, a compute credit allowance for the weekend, and a written review of your evaluation report.

Process steps

1Enquire; confirm your task and dataset access
2Receive pre-weekend setup instructions and notebooks
3Attend both days live online (6 hrs each)
4Receive written review of your evaluation report
Enquire About This Course
Fine-tuning diagram with loss curves
ML pipeline and monitoring dashboard
[course_3, pipeline, serve]

Machine Learning Engineering Programme

RM 4,600 · 20 weeks · ~12 hrs/week · part-time

Prerequisites

Professional software experience and familiarity with Python. A short technical conversation before enrolment makes sure the fit is right for both sides.

A five-month, part-time programme covering the engineering side of machine learning: data pipelines, versioning, feature stores, experiment tracking, reproducible training, model registries, packaging and serving, latency and cost profiling, canary releases, drift monitoring, and the operational habits that keep systems honest after launch.

Roughly 70 per cent of the time is project work. Each learner ships three reviewed projects and a capstone system that runs end to end. Reviews are done by practising engineers, line by line, in writing.

Syllabus areas

  • Data pipelines, versioning, feature stores
  • Experiment tracking, reproducible training
  • Model registries, packaging, serving
  • Latency / cost profiling, canary releases
  • Drift monitoring and post-launch operations

Not covered here

Research-style modelling or training large foundation models. Prompt engineering for hosted APIs (that's the API course). Data science and analysis work that doesn't involve production systems.

Included

Twice-weekly live sessions, recordings, three reviewed projects, capstone with written review, mentor office hours, cohort channel, and a repository you own.

Process steps

1Send enquiry; we arrange a short technical conversation
2Confirm place; receive cohort details and repo access
3Attend twice-weekly sessions, complete three reviewed projects
4Ship capstone system; receive written review
Enquire About This Programme
[compare, decision, matrix]

Which course fits your situation

Three different starting points. Use this to decide which one to enquire about.

Feature LM API Course Fine-Tuning Intensive ML Engineering Prog.
Duration 4 weeks 2 days 20 weeks
Price (MYR) 890 1,340 4,600
Live sessions
Written review 2 exercises eval report 3 projects + capstone
Technical prereq JS or Python dev PyTorch basics Software + Python
Compute included weekend credits
Best for Devs adding AI features Devs adapting models Devs owning ML systems
[standards, shared, protocols]

Standards that apply to all three courses

Privacy

Learner data is used only to operate the course. We don't share personal details with third parties or add people to mailing lists without explicit consent.

Syllabus currency

Materials are reviewed before each cohort starts. If the field has changed enough to matter, the syllabus is updated.

Enquiry response

We aim to reply to all enquiries within one working day. If a course isn't the right fit, we'll say so clearly rather than proceed with enrolment.

Company invoicing

Invoices can be issued to companies. Enquire with the billing entity name and details and we'll sort the paperwork.

No outcome claims

We describe a syllabus, a workload and a review process. We make no claim about employment, earnings or career outcomes of any kind.

Recordings included

All live sessions are recorded and available to enrolled learners for the duration of the cohort. No separate charge for recording access.

[pricing, MYR, stated]

Pricing

[short, 4wk, api]

LM API Course

RM 890

per cohort / 4 weeks

  • 4 live sessions + recordings
  • Reference repository
  • Evaluation harness (yours to keep)
  • Written feedback on 2 exercises
Enquire
[programme, 20wk, eng]

ML Engineering Programme

RM 4,600

per cohort / 20 weeks

  • Twice-weekly live sessions
  • 3 reviewed projects + capstone
  • Mentor office hours
  • Cohort channel + repo ownership
Enquire
[enquire, output, next]

Tell us what you're working with

Send a message with your technical background and what you're trying to do. We'll describe which course makes sense and why — or say honestly if none of them fit right now.

Send a Message