Services
Six categories of work.
Separated by the constraint that drives them, because that is what changes the engineering. Most engagements start in one and touch two.
LLM cloud
Model selection, gateways, routing and cost control across the major providers and open-weight models.
On-device & edge
Phones, laptops, vehicles and embedded targets — quantisation, runtimes, memory budgets and hybrid fallback.
Optimisation
Quantisation, distillation, adapters, speculative decoding and caching — each measured against a quality bar.
Deployment & serving
Serving stacks, GPU scheduling, model registries and release mechanics that survive a bad Friday.
Scaling & reliability
Capacity planning, autoscaling, queueing, SLOs and failover for workloads that cannot fall over.
Private & sovereign cloud
On-prem, VPC-isolated and air-gapped deployments, with the compliance evidence that makes them signable.
Choosing
Where to start.
If more than one of these describes you, start with the one that is costing you most this quarter. The diagnostic will find the rest.
| If this is true | Start here |
|---|---|
| Your token bill is growing faster than your usage | Optimisation |
| You are locked to one provider and it is starting to hurt | LLM cloud |
| The data cannot leave your boundary | Private & sovereign cloud |
| It must work on a phone, a vehicle, or with no network | On-device & edge |
| It works in a notebook and nowhere else | Deployment & serving |
| It falls over at peak, or you cannot forecast peak | Scaling & reliability |
Tell us what's breaking.
Bring a workload, a latency target, or a bill you can't explain. First conversation is a technical one — no discovery deck.