Voodoo AIVoodoo AI
Book a Consultation
Case Studies

Work examples without the theatre

These are representative engagement patterns: the kind of messy technical situations Voodoo AI is useful for, and the practical shape the work usually takes.

AI & Data

Supply Chain Forecasting Workflow

Representative AI & data pattern

Relevant if: you need AI or data systems to drive measurable operational improvement rather than isolated experimentation.

Situation

Planning teams had useful data, but forecasts, exceptions, and procurement decisions lived across disconnected tools and manual judgement calls.

Practical approach

Start with data readiness, workflow ownership, and exception handling. Then connect forecasting outputs into the operational systems where decisions are already made.

Read the Example

What improves

Cleaner demand-planning workflow
Fewer manual reconciliation loops
Clearer exception handling
Forecast outputs connected to daily decisions

Typical tools involved

KafkaPyTorchMLflowAirflowAWS SageMakerFastAPI
Cloud

Cloud Migration Without a Big-Bang Rewrite

Representative cloud modernisation pattern

Relevant if: you are modernising critical services without accepting unnecessary delivery risk or runaway cloud spend.

Situation

A critical estate needed modernising, but downtime, compliance, dependency risk, and unclear ownership made a simple migration plan unrealistic.

Practical approach

Design the target environment, map dependencies, move services in bounded batches, validate rollback paths, and introduce cost governance before spend becomes invisible.

Read the Example

What improves

Safer migration sequencing
Clearer service ownership
Better cost visibility
Lower operational uncertainty

Typical tools involved

AWSTerraformEKSGitHub ActionsArgoCDDatadog
Development

Node.js Platform Recovery

Representative platform architecture pattern

Relevant if: your platform is slowing product delivery, struggling under load, or carrying too much architecture debt.

Situation

A growing platform had become difficult to reason about: slow paths were unclear, releases felt risky, and new features touched too much of the system.

Practical approach

Identify the worst bottlenecks, make boundaries explicit, improve observability, and change the architecture in steps small enough for the team to absorb.

Read the Example

What improves

Faster diagnosis of production issues
Safer release boundaries
Clearer ownership of critical paths
More predictable delivery planning

Typical tools involved

Node.jsNestJSKafkaRedisPostgreSQLGrafana

Facing something similar?

If the details are different but the pattern feels familiar, we can help work out the first sensible move.

Book a Consultation