Forecasting
Demand, load, price, yield and failure forecasts, with honest error ranges.
Custom algorithm development.
Models and methods built around your data and your decision, not a generic tool adapted to fit. Delivered with the code and the evidence.
What is custom algorithm development?
Custom algorithm development is the design, build and testing of a model or method for one specific problem and one specific set of data. It differs from buying software in that the method is shaped by your constraints, your decision rules and your measure of success, and you own the result.
We build forecasting, optimisation, scheduling, anomaly detection, ranking, risk scoring and signal or image analysis, and we implement them inside your systems.
What we build.
Optimisation and scheduling
Routes, rosters, production plans and resource allocation under your real constraints.
Anomaly and fraud detection
Unusual behaviour in transactions, equipment and networks, surfaced for people to review.
Risk and scoring models
Transparent scores with documented inputs, so decisions can be explained and audited.
Ranking and recommendation
Prioritised lists for cases, leads, stock and content, tuned to your outcome.
Signals and images
Computer vision and signal processing for quality, inspection, imagery and sensors.
How it works.
- 1
Define the decision
The decision, the metric and today's baseline, agreed in writing.
- 2
Audit the data
Coverage, quality and rules, so the method fits what you actually have.
- 3
Build and compare
Candidate methods tested on held-back data against the baseline.
- 4
Deploy and hand over
Running in your environment, monitored, documented and owned by you.
What you receive.
Everything needed to run, check and change the algorithm without us.
- Source code and model files
- Documentation of the data, the method and its limits
- An evaluation report that compares the result with your baseline
- Deployment into your environment, including on-premise or offline
- A monitoring and retraining plan
- A handover session for your team
Scoped first, then a fixed quote
A free call and a short scoping step define the problem, the data and the measure of success. You then receive a fixed quote in writing. If a standard tool would do the job, we will say so. Prices in Australian dollars, excluding GST.
What is custom algorithm development?
It is the design and build of a model or method for your specific problem and data, rather than configuring a generic tool. Typical examples are forecasting, optimisation, scheduling, anomaly detection, ranking and risk scoring.
When do I need a custom algorithm instead of off-the-shelf software?
When your data, constraints or decision rules are specific enough that a generic tool gives poor results, or when the method itself is part of your advantage. If a standard tool meets your need, we will tell you and save you the cost.
What do I receive?
The source code and model files, documentation of the data and method, an evaluation report that compares the result with your baseline, a deployment into your environment, and a monitoring and retraining plan. You own all of it.
How is a custom algorithm priced?
We scope the problem in a free call and a short scoping step, then give you a fixed quote in writing. Prices are in Australian dollars and exclude GST.
How do you test that it works?
We agree a metric and a baseline first, hold back data the model has never seen, and report accuracy against the baseline before anything is deployed. After deployment we monitor the same metric.
Can it run inside our environment or offline?
Yes. We build for on-premise, private cloud, Australian cloud regions and fully offline deployment, so your data does not have to leave your control.