Three Engagements.
Each With a Defined Scope.
Forecasting systems, anomaly detection, and structured technical education — each described in detail below with process, deliverables, and fixed pricing in MYR.
Back to HomeHow We Approach Every Engagement
Every engagement begins with a written scope that states the deliverable, the data requirements, the evaluation criteria, and the timeline. No work starts without this document being agreed by both parties.
For build engagements, we hold a brief weekly written update throughout the project. These updates are factual: what was done, what comes next, and whether anything has arisen that might affect the timeline or scope.
Deliverables for build engagements are designed to be operated by the client's existing team. We document the system thoroughly, including the conditions under which it should not be trusted.
Define the business question, data requirements, evaluation criteria, and scope in writing.
Examine the data, identify preparation steps, and confirm the approach is viable.
Build the system, evaluate against the agreed baseline, report results honestly.
Deliver the system with full documentation; brief the operations team.
Forecasting System Build
An eight-to-ten week engagement for operating businesses with a clear, recurring forecasting need — demand, capacity, cash position — and the historical data required to support a quantitative approach. We examine the data, build and evaluate forecasting models against honest baselines, and deliver a system the firm's operations team can run themselves.
The work is plain about uncertainty; the deliverable presents forecasts with their uncertainty intervals, not as point estimates pretending to a precision they do not have.
- Evaluated against a simple, agreed baseline
- Uncertainty intervals included in all outputs
- Operations team briefed on maintenance
- Limitations document included
- Retraining process documented
Anomaly Detection for Operations
A focused build for firms whose operations produce streams of routine readings — manufacturing line outputs, transaction logs, infrastructure telemetry — and who would value a calm, well-tuned system to flag readings that warrant human attention.
We work with the operations team to define what counts as an anomaly worth raising, build the detection logic, and tune it against historical incidents. The deliverable is a working alerting system with a documented false-positive rate.
- Anomaly definition agreed with your operations team
- Tuned against historical incidents from your data
- False-positive rate stated in the deliverable
- Alert routing documented and configurable
- Operates on your existing data infrastructure
Reading Group on Current Methods
A six-session reading group conducted at the client's premises for senior technical staff who want a structured introduction to current methods in machine learning. Each session covers one topic — embedding spaces, fine-tuning, retrieval-augmented generation, evaluation methodology, deployment patterns, governance — with pre-reading provided and discussion led by a member of our team.
Suitable for in-house engineering leads and architects who need to evaluate ML vendor claims and architectural proposals without a specialised background in machine learning.
- Six sessions covering six distinct topics
- Pre-reading distributed before each session
- Discussion format, not lecture-style delivery
- Led by a practitioner, not a trainer
- Conducted at your premises (Peninsular MY)
Choosing the Right Engagement
The three engagements serve different needs. This table outlines the key characteristics to help you identify which fits your situation.
| Characteristic | Forecasting | Anomaly Detection | Reading Group |
|---|---|---|---|
| Requires historical data | |||
| Technical staff as primary audience | |||
| Produces a running software system | |||
| Conducted at client premises | Partial | Partial | |
| Best for forward-looking operational questions | |||
| Best for evaluating ML proposals & vendors |
Technical Protocols
Data Security
Client data is handled under a signed data agreement. It is not retained beyond the engagement and is not used in any other context.
Open Tooling
Build deliverables use open-source tooling where possible, avoiding vendor lock-in. We document all dependencies so they can be updated or replaced.
Written Evaluation
Every build engagement includes a written evaluation section comparing the system's performance to the agreed baseline on held-out data.
Limitations Documentation
Every deliverable includes a limitations section describing when the system's outputs should be treated with scepticism and what conditions require review.
Retraining Guidance
Build deliverables include a retraining procedure so the client's team can update the model when data distribution shifts or performance degrades.
Post-Handover Support
A thirty-day period of email-based support is included after each build engagement for questions arising from the handover documentation.
Fixed Fees in MYR
- Detection logic built & tuned
- False-positive rate documented
- Working alerting system
- Limitations section included
- 30-day post-handover support
- Baseline comparison & evaluation
- Uncertainty intervals in outputs
- Operations-ready system
- Full technical documentation
- Retraining procedure included
- 30-day post-handover support
- Six structured sessions
- Pre-reading per session
- Practitioner-led discussion
- At your premises
- Up to 12 participants
Not Sure Which Engagement Fits?
A brief conversation about your specific situation — the question you're trying to answer and the data you have available — is usually enough to determine which engagement, if any, makes sense. There is no charge for this conversation.
Book an Initial Discussion